Effect of Corporate Governance On Credit Ratings

ARTICLE IN PRESS
Journal of Accounting and Economics 42 (2006) 203–243

www.elsevier.com/locate/jae
The effects of corporate governance on firms’

credit ratings$
Hollis Ashbaugh-Skaifea, Daniel W. Collinsb,, Ryan LaFondc
a
University of Wisconsin, Madison, USA
b
University of Iowa, USA
c
Sloan School of Management, MIT, USA
Available online 16 June 2006
Abstract
We investigate whether firms with strong corporate governance benefit from higher credit ratings
relative to firms with weaker governance. We document, after controlling for firm-specific risk
characteristics, that credit ratings are negatively associated with the number of blockholders and
CEO power, and positively related to takeover defenses, accrual quality, earnings timeliness, board
independence, board stock ownership, and board expertise. We also provide evidence that CEOs of
firms with speculative-grade credit ratings are overcompensated to a greater degree than their
counterparts at firms with investment-grade ratings, thus providing one explanation for why some
firms operate with weak governance.
r 2006 Elsevier B.V. All rights reserved.
JEL classification: G30; G32; M41
Keywords: Corporate governance; Credit rating; Executive compensation
$
We would like to thank Sanjeev Bhojraj, Bob Bowen, Tom Dyckman, Paul Hribar, April Klein, S.P. Kothari,
Charles Lee, Mark Nelson, Kathy Petroni, Shiva Rajgopal, D. Shores, Joe Weber, Peter Wysocki, and seminar
participants at Cornell, University of Illinois Audit Symposium, Iowa State University, Journal of Accounting &
Economics Research Conference at Michigan, Lancaster University, London Business School, MIT, and the
University of Washington for helpful comments and suggestions. We especially thank Johannes Ledolter for
useful discussions on implementation and interpretation of ordered logit models and editor, Thomas Lys, for
detailed comments on our paper.
Corresponding author. Tel.: +1 319 335 0912; fax: +1 319 335 1956.
E-mail addresses: hashbaugh@bus.wisc.edu (H. Ashbaugh-Skaife), daniel-collins@uiowa.edu (D.W. Collins),
rzlafond@mit.edu (R. LaFond).
0165-4101/$ - see front matter r 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.jacceco.2006.02.003
ARTICLE IN PRESS
204 H. Ashbaugh-Skaife et al. / Journal of Accounting and Economics 42 (2006) 203–243
1. Introduction
We investigate whether firms that possess strong corporate governance benefit from
higher credit ratings relative to firms with weak governance. A firm’s credit rating reflects a
rating agency’s opinion of an entity’s overall creditworthiness and its capacity to satisfy its
financial obligations (Standard & Poor’s, 2002). Credit agencies are concerned with
governance because weak governance can impair a firm’s financial position and leave debt
stakeholders (hereafter referred to as bondholders) vulnerable to losses (FitchRatings,
2004). To structure our analysis, we adopt a framework developed by Standard & Poor’s
(2002) for assessing firms’ corporate governance structures and practices. Standard &
Poor’s (2002) framework focuses on four major components of governance: Ownership
Structure and Influence, Financial Stakeholder Rights and Relations, Financial
Transparency, and Board Structure and Processes. The governance attributes we examine
within each of these components are designed to increase the monitoring of management’s
actions to promote effective decision making, limit their opportunistic behavior and reduce
the information asymmetry between the firm and its external stakeholders. We investigate
what effect, if any, these governance features have on firms’ overall credit ratings.
Our analysis yields several key findings. First, we find variables that capture each of the
four major components of corporate governance enumerated above help explain overall
credit ratings after controlling for firm characteristics that prior research has shown to be
related to firms’ credit ratings. Specifically, we find that firms’ overall credit ratings are: (1)
negatively associated with the number of blockholders that own at least a 5% ownership in
the firm; (2) positively related to weaker shareholder rights in terms of takeover defenses;
(3) positively related to the quality of working capital accruals and the timeliness of
earnings; and (4) positively related to overall board independence, board stock ownership,
board expertise, and negatively related to CEO power on the board. To provide an
indication of the economic significance of our results, we find that moving from the lower
quartile to the upper quartile of the governance variables doubles a firm’s probability of
receiving an investment-grade credit rating—from 0.46 to 0.93.1 During the time frame of
our analysis, the average yield for firms with investment-grade debt with a 10-year
maturity was approximately 6.00%. In contrast, the average yield for firms with
speculative-grade debt with a 10-year maturity was approximately 14.0%. This 800-basis
point spread translates into an annual interest cost differential of $38.4 million for the
median firm in our sample with a speculative-grade credit rating.2
1
For purposes of this analysis, we hold the firm characteristic variables (ROA, LEV, SIZE, etc.) constant at the
mean values for the sample. For those governance attributes found to be positively (negatively) related to credit
ratings, our benchmark probability is determined by assigning governance values equal to the first (third) quartile
and then moving to the third (first) quartile value. For governance attributes measured as 0–1 dummy variables,
the benchmark probability is determined with the zero (one) value when the governance attribute is positively
(negatively) related to credit ratings.
2
To calculate this interest cost differential, we first determine the mean debt-to-asset ratio for our sample of
investment-grade firms (0.28). We then multiply this ratio times the total assets of each speculative-grade firm to
estimate the ‘‘as-if’’ capital structure if the speculative-grade firm was an investment-grade firm. Finally, we
multiply the estimated debt level times the 800-basis point spread to determine a speculative firm’s hypothetical
incremental interest cost due to weak governance. This additional interest can be viewed as a cost that
shareholders bear if the firm chooses not to implement stronger governance to preclude management from stealing
firm assets. However, this spread overstates the cost of weak governance from the shareholders viewpoint when
weaker governance allows wealth transfers from bondholders to shareholders to take place. To the extent that
ARTICLE IN PRESS
H. Ashbaugh-Skaife et al. / Journal of Accounting and Economics 42 (2006) 203–243 205
Our results suggest that weak governance can result in firms incurring higher debt
financing costs. So why are some firms willing to bear additional debt costs by not
practicing good governance? We approach this question by considering how CEOs can
extract rents from weak governance. One way CEOs can appropriate rents is through
excess compensation. To investigate this conjecture, we estimate CEO excess compensation
following the work of Core, Holthausen and Larcker (1999). We document that CEOs of
firms with weaker governance (greater CEO power or management entrenchment) receive
more excess compensation relative to the CEOs of firms with stronger governance (less
management entrenchment). Furthermore, we show that firms with speculative-grade debt
have a greater propensity to overcompensate their CEOs than do firms with investment-
grade debt. For firms with speculative-grade credit ratings, we then compare CEO excess
compensation to their share of additional debt costs that these firms bear due to weak
governance. We find that the median excess compensation far outweighs the CEO’s share
of the additional after-tax interest cost from having speculative-grade debt versus
investment-grade debt, thus providing one explanation for why all firms do not practice
good governance.
Our study makes several contributions to the extant literature on bond (credit) ratings
and corporate governance. Prior literature investigating firms’ credit ratings and debt costs
models the cost of debt as a function of issue characteristics and issuer risk attributes (see
e.g., Horrigan, 1966; Kaplan and Urwitz, 1979) while ignoring the governance mechanisms
that are put into place to safeguard the assets of the firm and ensure that bondholder
interests are well-served. We extend the traditional bond rating literature by incorporating
governance features designed to control agency conflicts between bondholders and
management and between bondholders and shareholders.
Sengupta (1998) and Bhojraj and Sengupta (2003) explore the effects of corporate
governance on debt ratings and cost of debt financing, but restrict their analysis to a
limited set of governance variables. Sengupta (1998) finds a negative relationship between
firms’ disclosure quality ratings and the cost of debt financing as reflected in realized yields
on new debt issues. Bhojraj and Sengupta (2003) find that firms with a higher percentage of
outside directors on the board and with greater institutional ownership enjoy lower bond
yields and higher ratings on their new debt issues. We extend these two studies by
evaluating a broader set of governance variables, thereby providing a more comprehensive
analysis of the relevance of corporate governance from the perspective of bondholders.
Much of the prior literature that investigates the effect of corporate governance focuses
on the value of corporate governance from the shareholder’s perspective (McConnell and
Serves, 1990; Yermack, 1996; Karpoff, Malatesta and Walkling, 1996; Gompers, Ishii and
Metrick, 2003). Our study focuses on the value of governance from the bondholders’
perspective because debt is a major source of capital for publicly traded firms.3 Thus, to the
extent that governance is an important determinant of credit ratings, it can have a
significant effect on firms’ external financing costs.
(footnote continued)
part of the interest rate spread between investment-grade and speculative-grade debt is due to bondholders price-
protecting against potential wealth transfers between themselves and stockholders, this spread represents an upper
bound on the cost of weak governance from the shareholder’s perspective.
3
For example, in calendar year 2000, U.S. firms obtained $944.8 billion dollars in debt financing whereas equity
financing totaled $311.9 billion dollars (U.S. Census Bureau, 2005).
ARTICLE IN PRESS
We investigate how various governance mechanisms that are intended to control agency
conflicts between management and all stakeholders impact credit ratings. In addition, our
study provides insights into the governance attributes that heighten or attenuate the
potential conflict between bondholders and shareholders. The interests of bondholders and
shareholders can diverge when there are differing stakes in firm performance and differing
views on management’s investment policies (FitchRatings, 2004). Gompers et al. (2003)
find that firms with stronger shareholder rights have higher share values and enjoy a lower
cost of equity capital. We find that firms with stronger shareholder rights have lower credit
ratings implying a higher cost of debt financing. Our study is one of the first to
demonstrate that governance mechanisms that benefit shareholders may do so at the
expense of bondholders.4 Thus, governance mechanisms designed to give more power to
shareholders can have wealth redistribution effects that leave bondholders worse off.
The remainder of the paper is organized as follows. Section 2 briefly describes the role of
governance in mitigating agency conflicts between bondholders and management and
between bondholders and stockholders. Section 3 sets forth the framework recently
adopted by Standard & Poor’s for evaluating the strength of firms’ corporate governance
mechanisms and develops empirical proxies to capture various elements within this
framework. Section 4 describes our sample, data sources, and variable measurements and
provides descriptive statistics. Section 5 presents the empirical models used to investigate
the relation between various corporate governance mechanisms and firms’ credit ratings
along with the main empirical results. In Section 6, we present evidence on CEO excess
compensation related to weak governance, address endogeneity issues, and conduct
sensitivity analyses. Section 7 concludes and offers suggestion for future research.
2. Why governance affects firms’ credit ratings
Firm credit ratings are determined by rating agencies’ assessments of the probability
distribution of future cash flows to bondholders, which in turn, depends on the future cash
flows to the firm. A firm’s creditworthiness is determined by assessing the likelihood that
its future cash flows will be sufficient to cover debt service costs and principal payments. As
the mean of the firm’s future cash flow distribution shifts downward or the variance of its
future cash flows increases, the likelihood of default increases and the firm’s credit rating
will decline.
Within the Jensen and Meckling (1976) agency theory framework, bondholders, and
more generally debt stakeholders, face two types of agency conflicts that can increase the
probability of default and, hence, reduce the value of their claims. The first is the conflict
between management and all external stakeholders—both bondholders and shareholders.
Separation of ownership and control in corporate organizations leads to information
asymmetry problems between external stakeholders and managers. Information asym-
metry creates a moral hazard problem when managers have incentives to pursue their own
interests at the expense of external stakeholders. Self-interested managerial behavior can
take several forms including shirking, consumption of perquisites, overcompensation, and
empire building, all of which increase the agency risk faced by external stakeholders and
4
Our results are consistent with a concurrent study by Klock, Mansi and Maxwell (2004) who find that firms
with stronger anti-takeover provisions (weaker shareholder rights) enjoy a lower cost of debt financing relative to
firms with weaker anti-takeover provisions.
ARTICLE IN PRESS
decrease the expected value of the cash flows to the firm and its external stakeholders. As
the firm’s expected cash flows decline, the default risk of bondholders increases leading to
lower credit ratings.
The second agency conflict faced by bondholders is the conflict with shareholders.
Shareholders in levered firms have incentives to undertake actions that can transfer wealth
from bondholders to themselves. This wealth transfer can take several forms that affect the
mean and the variance of the firm’s future cash flows. For example, if shareholders demand
direct payouts of firm assets (dividends or share repurchases) as opposed to supporting
manager’s investments in positive net present value projects, then the mean of a firm’s
future cash flow distribution will be lower. The reduction in a firm’s expected future cash
flows increases bondholders’ default risk. Likewise, if shareholders influence managers to
invest in riskier projects that increase the variance of a firm’s future cash flows,
bondholders face greater default risk. In both examples, bondholders bear greater risk that
their fixed contractual claims on the firm’s cash flows will not be paid while shareholders
potentially are better off.
We hypothesize that governance features impact credit ratings by controlling agency
costs that result from conflicts between managers and all external stakeholders as well as
between bondholders and shareholders. Many of the governance features we examine are
designed to reduce the agency conflict between managers and all stakeholders. Governance
mechanisms that provide independent monitoring of management promote effective
managerial decision making that increases firm value and guard against opportunistic
management behavior that decreases firm value. Governance mechanisms that promote
better managerial decision making and limit opportunistic management behavior benefit
all stakeholders. Conversely, we posit that if governance is weak, the firm’s distribution of
future cash flows will shift downward relative to what it would be with more effective
governance. This increases the likelihood of default resulting in a lower credit rating. For
convenience, we refer to the role that governance plays in mitigating the agency conflicts
between management and all stakeholders as the ‘‘management disciplining’’ hypothesis.
Shareholder and bondholder interests are generally aligned when better monitoring of
management occurs. However, certain elements of corporate governance have a more
ambiguous impact on bondholders (FitchRatings, 2004). For example, some features of
governance can place greater power in the hands of shareholders (or selected subsets of
shareholders) who can assert their influence to obtain preferential treatment at the expense
of other stakeholders (e.g., greenmail or targeted share repurchases (Dann and DeAngelo,
1983)). Alternatively, shareholders can use their voting power to encourage management
to undertake risky investments or engage in ownership changes that can harm bondholder
interests. Taking on riskier projects increases the likelihood of default, resulting in lower
credit ratings. Some of the governance features we consider below (e.g., shareholder rights)
have the potential for affecting wealth transfers between bondholders and shareholders.
Hence, while beneficial from the shareholders, perspective, certain governance features
potentially can be harmful to bondholders.5 Or, alternatively, governance features that
weaken shareholder rights may actually be viewed positively from the bondholders’
5
For example, shareholders will only approve mergers or acquisitions that serve their interests. But bondholders
do not always benefit under all takeover scenarios (see Asquith and Wizman, 1990; Warga and Welch, 1993). So
giving shareholders greater power to determine ownership changes may well be viewed as an additional risk factor
by bondholders and rating agencies.
ARTICLE IN PRESS
perspective. We refer to this as the ‘‘wealth redistribution’’ hypothesis regarding how

governance impacts credit ratings.
In sum, the governance variables introduced in the next section proxy not only for the
agency conflicts between external stakeholders and management, but also potential
conflicts between bondholders and shareholders that can result in wealth transfer effects
between these two stakeholder groups.
3. Corporate governance attributes
Prior studies on corporate governance tend to focus on one attribute of governance, e.g.,
board independence, as opposed to studying a broad set of governance attributes intended
to protect stakeholders’ claims to firms’ resources (Sengupta 1998; Bhojraj and Sengupta,
2003; Hermalin and Weisbach, 1991; Bhagat and Black, 2000). A limitation of this
research is that some governance attributes may complement each other in protecting
stakeholders’ claims whereas other governance attributes may serve as substitutes. As a
result, inferences drawn from studying one attribute of governance may be subject to an
omitted variables problem. In 2002, Standard & Poor’s developed a comprehensive
framework for evaluating corporate governance that is based on four governance
components; ownership structure and influence, financial stakeholders rights and relations,
financial transparency and disclosure, and board structure and processes (Standard &
Poor, 2002). We use Standard & Poor’s framework, along with prior literature, to identify
governance attributes that potentially affect firms’ credit ratings. The four dimensions of
governance and the empirical proxies to capture the major attributes of governance within
each dimension are discussed below.
3.1. Ownership structure and influence
Typically, corporate governance is viewed from the perspective that publicly traded
firms have dispersed shareholders who demand governance to protect their residual claims.
Governance mechanisms that monitor management actions and limit their opportunistic
behavior protect the interests of residual claimants (shareholders) and the interests of
bondholders as well. At times, however, the interests of shareholders and bondholders can
diverge. For example, shareholders with significant ownership positions can exercise their
influence to force management to take on more risky investments where shareholders as a
group receive the benefits of successful outcomes, but bondholders bear a disproportionate
share of the failures. Ownership structure and the influence that certain shareholders exert
on management play a key role in determining the potential for wealth transfers between
bondholders and stockholders, and are important elements of corporate governance.
Jensen (1993) and Shleifer and Vishny (1997) argue that blockholders or institutional
investors that hold large equity positions in a company are important to a well-functioning
governance system because they have the financial interest and independence to view firm
management and policies in an unbiased way, and they have the voting power to put
pressure on management if they observe self-serving behavior. Consistent with this view,
Gordon and Pound (1993) find that the structure of share ownership significantly
influences voting outcomes on shareholder-sponsored proposals to change corporate
governance structures. Outside blockholders and institutions (when institutional holdings
are relatively concentrated) tend to align with the proposal sponsor, while insiders and
ARTICLE IN PRESS
outside directors who hold significant stock positions tend to align strategically with
management, who often oppose the shareholder-sponsored proposals. Nesbitt (1994) finds
that firms targeted by the California Public Employees’ Retirement System (CalPERS)
experience positive long-run stock returns, and Opler and Sokobin (1997) find that firms
experience above-market performance the year after being targeted by the Council of
Institutional Investors. These results suggest that blockholders and active institutional
shareholders lead to more efficient monitoring of management and less managerial
opportunistic behavior, which benefits all stakeholders. We characterize this as the
‘‘management disciplining’’ role of governance and we predict a positive relation between
our institutional ownership and blockholder measures (described below) and credit ratings.
A competing view in the literature suggests that concentrated ownership allows these
shareholders to exercise undue influence over management to secure benefits that are
detrimental to minority shareholders and bondholders (Shleifer and Vishny, 1997; Bhojraj
and Sengupta, 2003). Examples include greenmail and targeted share repurchases (Dann
and DeAngelo, 1983). Under the ‘‘wealth redistribution’’ hypothesis, as the percentage of
shares held by institutions increases or as the number of blockholders increases, the
likelihood of these shareholders using their influence to affect wealth transfers from
bondholders increases, suggesting a negative relation between credit ratings and our
institutional ownership and blockholder variables.
We capture the ownership effects of governance with three variables. BLOCK is the
number of outside blockholders that own 5% or more of a firm’s outstanding voting
stock.6 %INST measures the percentage of shares held by institutional investors. The
relation between these two ownership structure variables and firm credit ratings depends
on whether these ownership concentrations, on average, are beneficial to bondholders
(management disciplining hypothesis) or further the interests of shareholders at the
expense of bondholders (wealth redistribution hypothesis). Because we have no way of
predicting, a priori, which effect is likely to dominate, we leave the prediction on these two
variables unsigned. The third variable, %INSIDE, is the percentage of shares held by
officers or directors.7 We predict that %INSIDE will be negatively related to RATING
under the assumption that insiders will use their voting power to expropriate firm resources
for their personal benefit or resist shareholder-sponsored proposals to increase the
monitoring of their actions (Gordon and Pound, 1993), both of which are likely to lead to
greater agency risks for bondholders. In addition, we predict a negative relation between
%INSIDE and RATING because increasing insider ownership results in stronger
incentives for officers and managers, as residual claimants, to invest in projects that have
very high returns when successful but low probabilities of success (Jensen and Meckling,
1976)—that is, projects that increase bondholders’ risk due to the differential payoff
structure between bondholders and shareholders (wealth redistribution hypothesis).
6
An alternative construct to capture the power of significant ownership is to use the percentage of shares held by
the largest shareholder. Board Analyst has a variable labeled dominant shareholder, which reflects whether the
firm has a shareholder holding a significant proportion of shares. There are 146 of our 894 sample firms that have
a dominant shareholder owning more than 10% of the outstanding shares. When we estimate our model that
includes a dummy variable that captures firms that have a dominant shareholder, we find the coefficient on the
dominant shareholder variable to be insignificant.
7
Although this measure includes holdings by both officers and directors, the vast majority of %INSIDE is made
up of officer shareholdings. Thus, we expect this measure to largely proxy for managements’ self-interests rather
than board member incentives to monitor the actions of management.
ARTICLE IN PRESS
3.2. Financial stakeholder rights and relations
Financial stakeholder rights reflect the balance of power between stakeholders

(bondholders and shareholders) and management. A key element of this dimension of
corporate governance is whether the firm maintains a level playing field for corporate
control and whether it is open to changes in management and ownership that provide
increased shareholder value. Gompers et al. (2003) construct an index based on 24
governance provisions, referred to as the G_SCORE, to measure the power-sharing
relationship between investors and management. The 24 provisions are classified into five
categories of management power: (1) tactics for delaying hostile bids; (2) voting rights; (3)
director/officer protection; (4) other takeover defenses; and (5) state takeover laws. Each
firm’s G_SCORE is the sum of points, where one point is awarded for the presence of each
governance provision. Thus, a higher G_SCORE indicates lower shareholder rights and
greater management power.8 We use the Gompers et al. (2003) G_SCORE metric to proxy
for the stakeholder rights component of governance.
Takeover defenses and other restrictions of shareholder rights like staggered terms of
directors, supermajority voting requirements for approval of mergers and ownership
changes, and limits on shareholders’ ability to meet and act places more power in the hands
of management vis-à-vis shareholders and can make it difficult to remove entrenched
management that is acting opportunistically. Consistent with the management disciplining
hypothesis, governance mechanisms tilted in favor of opportunistic management can lower
overall firm value, resulting in losses to both shareholders and bondholders. This line of
reasoning leads to a predicted negative relation between G_SCORE and credit ratings.
That is, a firm with higher G_SCORE (less shareholder power) is expected to exhibit a
lower credit rating because it is more difficult or costly to remove management that is
acting opportunistically when managers have greater power vis-à-vis shareholders.
Alternatively, we may find a negative relation between G_SCORE and credit ratings
because firms with stronger shareholder rights (lower G_SCORE) are likely to provide
better monitoring and control over management leading to more effective and efficient
managerial decision making, which in turn leads to better overall firm credit worthiness
and higher credit ratings.9
Giving greater power to shareholders to determine changes in management as well as
ownership control does not necessarily make bondholders better off (FitchRatings, 2004).
For example, Asquith and Wizman (1990) and Warga and Welch (1993) find that pre-
buyout bondholders suffer significant wealth losses in leveraged buyouts. These results
suggest that bondholders do not always benefit under all takeover scenarios. Therefore,
8
Using a sample of 1500 firms during the 1990s, Gompers et al. (2003) find that taking a long position in firms
with the strongest shareholder rights and a short position in firms with the weakest shareholder rights yields an
average abnormal return of 8.5% per year.
9
Consistent with this conjecture, Gompers et al. (2003) find that firms with lower G_SCORES have higher firm
values, higher profits, higher sales growth, lower capital expenditures, and lower corporate acquisitions. These
factors should lower firms’ credit risk leading to higher credit ratings. That is, firms with stronger shareholder
rights are expected to have higher credit ratings because of lower risk characteristics. Controlling for these
additional risk characteristics does not change the sign or significance of the coefficient on G_SCORE reported
below.
ARTICLE IN PRESS
governance mechanisms that place greater power in the hands of shareholders may
actually be viewed negatively by bondholders and credit rating agencies because it
increases the likelihood of ownership changes that can transfer wealth from bondholders
to stockholders. This line of reasoning predicts a positive relation between G_SCORES
and credit ratings. That is, firms with higher G_SCORES (weaker shareholder rights)
are expected to have higher credit ratings. Consistent with this conjecture, Klock
et al. (2004) find that firms with stronger anti-takeover provisions (weaker shareholder
rights) have a lower cost of debt financing relative to firms with weaker anti-takeover
provisions.
Given the mixed evidence on whether greater shareholder power translates into benefits
for bondholders (management disciplining hypothesis) or greater risk due to potential
wealth transfers between bondholders and shareholders, we make no directional prediction
for the G_SCORE variable.
3.3. Financial transparency
Transparent financial reporting is critical to reducing the information asymmetry

between the firm and its capital suppliers. We posit that greater financial transparency
facilitates the monitoring of managements’ actions and makes it less likely that
management will act opportunistically. Sengupta (1998) conjectures that firms with
more timely and informative disclosures are perceived to have a lower likelihood of
withholding value-relevant unfavorable information. As a result, they are expected to be
charged a lower risk premium by creditors. Consistent with this prediction, he finds
that firms with higher AIMR disclosure ratings enjoy a lower effective interest cost of
issuing new debt. As AIMR disclosure ratings are no longer available, we use the quality
of firms’ working capital accruals, WCAQ, and the timeliness of firms’ earnings,
TIMELINESS, to capture the transparency of firms’ financial reporting. We describe
the measurement of WCAQ and TIMELINESS in detail in Section 4. Briefly, WCAQ
is based on the standard deviation of the firm-specific residual from regressing
working capital accruals on past, contemporaneous, and future operating cash flows,
where smaller residuals reflect a better mapping of working capital accruals to cash
flows (Dechow and Dichev, 2002). TIMELINESS is the squared residual from regress-
ing returns on earnings allowing for separate intercepts and slopes for profit and
loss firms (Gu, 2002). Earnings that better articulate with market returns are deemed
to be more transparent and timely in that they better reflect the economic events
that are priced by the market. A high-squared residual indicates that earnings are
less transparent/timely. To facilitate the interpretation of these variables, we multi-
ply the variables by negative one and predict a positive relation with firms’ credit
ratings.
The reliability of financial information is due, in part, to the quality and integrity of the
audit process. To proxy for the quality and integrity of the audit process, we use three
measures: (1) the total fees (audit plus non-audit) charged to the client firm divided by the
total revenues of the audit firm (TOTFEES); (2) %AUD_IND is the percentage of the
audit committee made up of outside independent directors; and (3) a dummy variable,
FIN_EXPERT coded one if the firm’s audit committee has at least one individual deemed
to be a ‘‘financial expert,’’ and zero otherwise. Using the attributes of a financial expert set
forth by the Securities and Exchange Commission (SEC, 2003) this variable is coded one if
ARTICLE IN PRESS
the audit committee has an outside independent director that is a CPA or who has
experience as a chief financial officer of another company.10
DeAngelo (1981) posits that auditor independence is threatened as the economic bond
between the auditor and client firm increases. Concern over economic bonding between the
client firm and its auditor was the major impetus behind the restrictions that Sarbanes-
Oxley placed on the kinds of non-audit services that auditing firms can perform for their
clients (U.S. Congress, 2002). However, the evidence on whether economic bonding
between the audit firm and its client impairs auditor independence as proxied by abnormal
accruals is mixed. While Frankel, Johnson and Nelson (2002) find evidence consistent with
economic bonding impairing auditor independence, Ashbaugh, LaFond and Mayhew
(2003), Chung and Kallapur (2003) and DeFond, Raghunandan and Subramanyam (2002)
do not find evidence of independence impairment. We use our TOTFEES variable to
measure the economic bonding between the audit firm and its client.11 If credit rating
agencies perceive that auditor independence, and thus the quality of financial statements is
impaired due to economic bonding, then we expect a negative relation between this
variable and firms’ credit ratings. However, if credit rating agencies perceive that the
economic bond between auditors and their audit clients do not threaten the quality of
firms’ financial reporting, as some of the studies noted above indicate, then we expect to
find no relation between TOTFEES and credit ratings.
The conventional wisdom is that audit committees more effectively carry out their
oversight of the financial reporting process if they include a strong base of independent
outside directors. Klein (2002) provides evidence to support this assertion. She finds a
negative relation between audit committee independence and abnormal accruals. She also
reports that as audit committee independence declines, abnormal accruals increase. To the
extent that better monitoring of the financial reporting process leads to less managerial
opportunism and better financial transparency, this should lead to lower default risk for
bondholders. Accordingly, under the management disciplining hypothesis, we predict a
positive relation between %AUD_IND and credit rating.
Several recent studies investigate whether including independent directors with financial
expertise on the audit committee is likely to improve the quality and integrity of a firm’s
financial statements. Klein (2003) finds that placing independent directors with
professional certifications in accounting or law or prior banking experience on the audit
committee reduces earnings management. Agrawal and Chadha (2003) provide
corroborating evidence and find that companies whose boards or audit committees have
an independent director with a background in accounting or finance face a lower
10
The SEC recently adopted this provision of the Sarbanes-Oxley Act (SEC, 2003) and defined an ‘‘audit
committee financial expert’’ to mean a person who has the following attributes:
(1) An understanding of financial statements and generally accepted accounting principles.
(2) An ability to assess the general application of such principles in connection with the accounting for estimates,
accruals, and reserves.
(3) Experience preparing, auditing, analyzing, or evaluating financial statements that present a breadth and level
of complexity of accounting issues that are generally comparable to the breadth and complexity of issues that
can reasonably be expected to be raised by the registrant’s financial statements, or experience actively
supervising one or more persons engaged in such activities.
(4) An understanding of internal controls and procedures for financial reporting.
(5) An understanding of audit committee functions.
11
We consider alternative ways of measuring this construct in the ‘‘sensitivity analysis’’ section below.
ARTICLE IN PRESS
probability of earnings restatements. DeFond, Hann and Xu (2004) and Davidson, Xie
and Xu (2004) find a positive market reaction to firms’ announcements that an accounting
financial expert has been appointed to the audit committee. To the extent that having a
financial expert on the audit committee improves board effectiveness and enhances the
integrity of the financial reporting process, we predict this will lead to less managerial
opportunism. Accordingly, under the management disciplining hypothesis, we predict a
positive relation between FIN_EXPERT and credit ratings.
3.4. Board structure and processes
This component of corporate governance deals with such things as: (1) board size and
composition in terms of proportion of inside, outside, and affiliated directors; (2) board
leadership and committee structure; (3) how competent and engaged board members are;
(4) whether there are a sufficient number of outside independent directors on the board
that represent the interests of all stakeholders, and how those members are distributed
across the various committees; and (5) whether board members are remunerated and
motivated in ways that ensure the long-term success of the company.
The first three elements address the board’s role and ability to provide independent
oversight of management performance and hold management accountable to stakeholders
for its actions. Boards often delegate oversight of key functions or decision making to
standing committees—e.g., audit, compensation, nominating or governance, finance and
investment. These committees, made up of subsets of board members, meet separately
from the full board and generally have specific, narrowly defined functions.
Prior research generally posits a positive relation between board and committee
independence and firm performance. Better firm performance should benefit all
stakeholders leading to higher credit ratings. However, research findings on the relation
between board and committee composition and overall firm performance are mixed.
Baysinger and Butler (1985) and Hermalin and Weisbach (1991) find no significant
association between the percentage of outsiders on the board and same-year measures of
corporate performance. Bhagat and Black (2000) find no relation between overall board
independence and four measures of firm performance (Tobins’ Q, return-on-assets,
market-adjusted stock returns and ratio of sales-to-assets) measured over a 3-year window.
Agrawal and Knoeber (1996) investigate the relation between firm performance (Tobin’s
Q) and seven control mechanisms including percentage of non-officer board members.
Using a simultaneous equations framework to control for the interdependence among the
various control mechanisms, Agrawal and Knoeber find a significant negative relation
between outside membership on the board and firm performance, leading them to
conclude that boards seem to have too many outsiders.
Klein (1998) extends the previous research on board composition and firm performance
by examining the relation between the composition of the overall board and of various
committees and firm performance. Consistent with prior evidence, Klein finds no
association between firm performance and overall board composition. Moreover, she
finds no association between the level of independence on audit, compensation and
nominating committees, and firm performance. Interestingly, she does find a significant
positive association between the percentage of inside directors on finance and investment
committees and accounting and stock market performance measures. One explanation for
this result is that inside board members bring specialized institutional and industry-specific
ARTICLE IN PRESS
knowledge to the table that helps these committees select long-term investment and
financing strategies that enhance firm value. Thus, inside board members appear to serve a
useful role in overall corporate governance if strategically placed on committees that have
more of an operating focus than a monitoring focus.
Finally, and more germane to bondholder interests, Bhojraj and Sengupta (2003) posit
that firms with a greater proportion of outside directors on the board provide better
monitoring of management actions. Thus, bondholders face less agency risks vis-à-vis
management. They posit that this, in turn, will lead to higher bond ratings and lower debt
yields. Consistent with this conjecture, they find that firms with a higher proportion of
non-officer directors enjoy lower bond yields and higher ratings on new bond issues.
Based on the literature reviewed above, we use %BRD_IND to measure the percentage
of board made up of independent outside (nonaffiliated) directors and predict a positive
relation between %BRD_IND and credit ratings.
Imhoff (2003) argues that board governance is severely compromised when the current
or former CEO of the company also serves as chairman of the board. This is because the
board chairman frequently sets the board’s agenda and, therefore, controls issues brought
before the board. Moreover, CEOs that serve as board chair frequently have significant
influence on the slate of candidates for board seats, thereby increasing the risk that new
board appointees will not be independent of management even though they are
‘‘outsiders’’. CEOs can also exert significant influence over the board through the
committees they serve on. We use CEOPOWER as a composite measure of the influence
that the CEO exercises over the board. A firm receives one point if the CEO is chairman of
the board and one point for each committee that the CEO serves on. We posit that CEO
power on the board is likely to reduce the board’s disciplining of opportunistic
management. Accordingly, under the management disciplining hypothesis, we predict
CEOPOWER will be negatively related to credit ratings.
Ceteris paribus, we expect that boards comprised of members who are more competent
or knowledgeable will do a better job of monitoring the activities of management and
make better decisions leading to less default risk. Similar to Klein (1998), we measure
board competency or expertise by the percentage of outside board members that sit on
boards of other companies (%BRD_EXPERT). We predict a positive relation between
this variable and credit ratings.
Board compensation is another element of the ‘‘Board Structure and Process’’
component of governance. Key issues are whether board members are remunerated and
motivated in ways that ensure the long-term success of the company. Jensen (1993) argues
that boards with greater ownership in the firm are more likely to do a better job of
monitoring management and fulfilling their fiduciary responsibilities. Consistent with this
conjecture, Yermack (2003) finds that director stock and option awards are positively
related to firms’ investment opportunities and subsequent firm performance. Yermack
shows that tying directors’ pay more closely to stock performance through the use of
options and other equity awards generally leads to increased performance. We use
%BRD_STOCK to measure the percentage of directors that hold stock in the company
and predict a positive relation between this variable and credit ratings.
Recently, the SEC endorsed the proposals of the NYSE and NASDAQ that firms adopt
a formal governance policy that outlines the roles and responsibilities of directors and
establishes an explicit code of business conduct and ethics for directors (SEC, 2003). We
expect that having such a formal governance policy places increased responsibility on
ARTICLE IN PRESS
board members and increases their legal liability leading to greater attentiveness on the
part of board members. Greater board attentiveness should lead to better monitoring of
management which, in turn, should lead to less management opportunism. We code
GOVERNANCE_POLICY with a one if a firm has a formal governance policy, and zero
otherwise, and predict a positive relation between this variable and credit ratings under the
management disciplining hypothesis.
Finally, we use %FINCOM_INSIDE to measure the percentage of insiders on finance
committees.12 Based on Klein’s (1998) results that having insiders on finance and/or
investment committees improves firm performance, we expect this committee structure
variable to be positively related to credit ratings since improved firm performance is
expected to improve a firm’s creditworthiness.
4. Sample, variables, and descriptive statistics
4.1. Sample and data sources
Data for this study are compiled from four sources:
Governance measures, audit/non-audit fees. and share ownership data—Board Analyst

database and firm proxy statements.
G_SCORES—Gompers et al. (2003).
Credit ratings and accounting variables—Standard & Poor’s Compustat.
Stock return data- CRSP.
We obtain the majority of the corporate governance measures from the Board Analyst
database compiled by The Corporate Library, an independent research firm that provides
data and analysis of corporate governance issues.13 This database contains detailed
governance, audit fee data, and stock ownership data (including institutional and inside
ownership) for over 2000 U.S. companies and profiles on over 22,000 individual directors.
The data used in our primary analysis are from the 2003 proxy season covering the board
and committee structures of firms for the 2002 fiscal year.
G_SCORES that measure the power-sharing relationship between investors and
management were obtained from Gompers et al. (2003). These G_SCORES are available
for approximately 1500 firms and are based on the incidence of 24 governance provisions
related to shareholder rights and takeover defenses found in 2002 proxy statements.
For firm credit ratings (RATING), we use the long-term issuer credit ratings compiled
by Standard & Poor’s and reported on Compustat (data item 280). The ratings range from
AAA (highest rating) to D (lowest rating—debt in payment default). These ratings reflect
S&P’s assessment of the creditworthiness of the obligor with respect to its senior debt
obligations. For purposes of our analysis, the multiple ratings are collapsed into seven
categories according to the schedule provided in Table 1. To facilitate the discussion of the
12
For those firms without finance committees, we used the percentage of insiders on the overall board for this
variable because, in the absence of a finance committee, the overall board would be charged with voting on
financial policy matters (see Klein, 1998 for similar treatment).
13
Board Analyst does not provide information on finance and investment committees. This information was
hand-gathered from 2003 proxy statements.
ARTICLE IN PRESS
Table 1
Credit rating classifications
S&P Debt rating Compustat data280 Assigned RATING Grade

score
AAA 2 7 Investment
AA+ 4 6 Investment
AA 5 6 Investment
AA 6 6 Investment
A+ 7 5 Investment
A 8 5 Investment
A 9 5 Investment
BBB+ 10 4 Investment
BBB 11 4 Investment
BBB 12 4 Investment
BB+ 13 3 Speculative
BB 14 3 Speculative
BB 15 3 Speculative
B+ 16 2 Speculative
B 17 2 Speculative
B 18 2 Speculative
CCC+ 19 1 Speculative
CCC or CC 20,23 1 Speculative
C 21,24 1 Speculative
D or SD 27,29,90 1 Speculative
Firm credit ratings (RATING) are the long-term issuer credit ratings compiled by Standard & Poor’s and
reported on Compustat (data item 280). The ratings range from AAA (highest rating) to D (lowest rating—debt in
payment default). These ratings reflect S&P’s assessment of the creditworthiness of the obligor with respect to its
senior debt obligations. For purposes of our analysis, the multiple ratings are collapsed into seven categories
according to the schedule provided above. S&P classifies ratings below BBBas speculative.
economic significance of our results, we also estimate our logistic regression model using a
two-category classification scheme—investment grade and speculative grade. The assign-
ment of the credit rating groups into these two classifications are also shown in Table 1.
Table 2, Panel A summarizes the sample selection procedure and number of firms lost
because of minimum data requirements from each data source. Essentially, our final
sample for the credit rating analysis is determined by the intersection of firms for which
required data are available on the four data sources noted above.14
Panel B of Table 2 provides details on board and committee composition for our sample
firms. Out of 894 sample firms, all have audit committees, 99.6% (890) have compensation
committees, 90.9% (813) have nominating committees, but only 27.1% (242) have finance
committees. The average board (committee) size is 10 (4) directors. The incidence of firms
having at least one insider on a committee ranges from 0.7% (6/894) for audit committees
to 30.2% (73/242) for finance committees. The much higher incidence of insiders on
finance committees presumably reflects the fact that insiders bring valuable institutional-
14
In general, our sample firms are larger than the average firm on Compustat with sample means of assets, sales,
market value of equity, and long-term debt (in millions) of $20,765, $7,502, $8,982, and $4,021, respectively. In
addition, 84%, 15%, and 1% of the sample firms’ shares trade on the New York Stock Exchange, NASDAQ, and
the American Stock Exchange, respectively.
ARTICLE IN PRESS
Table 2
Sample details
Panel A: Sample construction (fiscal year 2002)

Number of firms Firms lost
Number of firms in the main Board analyst dataset 2050 0

Number of firms having complete governance data from Board Analyst 1867 183
Number of firms having necessary Compustat data for ratings model 1693 174
Number of firms having G_SCORES 1392 301
Number of firms having credit ratings on Compustat 894 498
Panel B: Sample firms’ board composition

Breakdown of inside, outside and affiliate directors by board and by committees for sample firms in fiscal year
2002
Board Audit Compensation Nominating Finance
committee committee committee committee
Number of firms having a 894 894 890 813 242

board (committee)
Average number of 10.05 3.93 3.72 3.97 4.31
directors on board
(committee)
Average number of insider 1.67 0.01 0.02 0.06 0.39
directors on board
(committee)
Average number of outside 7.02 3.63 3.37 3.46 3.42
directors on board
(committee)
Number of firms having at 885 6 17 42 73
least one insider on the
board (committee)
Number of firms having 652 3 10 31 65
the CEO as chairman of
the board (committee)
Notes: This table presents the sample selection and data requirements. Panel B presents the size of the board and
committees for the sample of firms for which the specific committee exists. The difference between the average
number of directors on the board (committee) and number of insiders and outsiders on the board (committee) is
due to affiliated directors.
and industry-specific knowledge and expertise to this committee (Klein, 1998). Roughly
73% of our sample firms have CEOs that serve as Chairman of the Board, and the more
common committees that CEOs serve on are the nominating and finance committees.
4.2. Independent variables
4.2.1. Corporate governance measures

The variables identified in Section 3 that we use to capture key governance attributes
within the S&P framework are summarized in Panel A of Table 3 along with their
predicted relation with RATING. Except for our measures of financial transparency, the
variable measurements were described in detail in Section 3 when introduced, so we do not
take time to repeat them here.
ARTICLE IN PRESS
Table 3
Variable definitions
Variables Predicted Definitions and data source

sign
Ownership structure and influence

BLOCK ? Number of block holders, where block is defined at the 5% ownership
level (source Compact Disclosure).
%INST ? % of shares held by institutional investors (source Board Analyst).
%INSIDE % of shares held by insiders (officers and directors) (source Board
Analyst).
Financial stakeholder rights and relations.

G_SCORE ? Shareholder rights governance score (source Gompers et al., 2003)).
Financial transparency
WCAQ + Negative one times the standard deviation of the firm-specific residual
from the prior 3–5 years, where residuals are from the following
cross-sectional estimation of Dechow and Dichev’s (2002) model:
WCAt ¼ b0 þ b1 CFOt1 þ b2 CFOt þ b3 CFOtþ1 þ where
regressions are estimated by three-, two-, or one-digit SIC codes
conditional on having at least 10 firms in each SIC group.
WCA ¼ working capital accruals ((Compustat # 302+Compustat
# 303+Compustat # 304+Compustat # 305+Compustat # 307));
CFO ¼ cash flow from operations (Compustat # 308); all variables
are scaled by average total assets (Compustat # 6).
TIMELINESS + Negative one times the squared residual from the following regression
RET ¼ b0 þ b1 NIBE þ b2 LOSS þ b3 NIBE LOSS þ b4 DNIBE þ
where the regression is estimated by three-, two-, or one-digit SIC
codes conditional on having at least 10 firms in each SIC group.
RET ¼ the market-adjusted return over the fiscal year (from CRSP);
NIBE ¼ net income before extraordinary items (Compustat # 18)
scaled by beginning of period market value of equity (Compustat #
25*Compustat # 199); LOSS ¼ one if NIBE is negative, zero
otherwise; DNIBE ¼ the change in net income before extraordinary
items (Compustat # 18) scaled by beginning of period market value of
equity (Compustat # 18*Compustat # 199).
TOTFEES Total fees paid by the firm to its auditor divided by total revenues of
the audit firm multiplied by 100 (source firms’ proxy statements,
Accounting Today Top 100 firms and D&B’s Million Dollar
Database).
%AUD_IND + % of audit committee made up of independent directors (source
Board Analyst).
FIN_EXPERT + One if the firm has an independent financial expert on the audit
committee, where financial expert is defined as an audit committee
member being a CFO or having a CPA, zero otherwise (source Board
Analyst).
Board structure and processes

%BRD_IND + % of independent directors on the board (source Board Analyst).
CEOPOWER Composite score representing the power of the CEO where a firm
receives one point if the CEO is the chairman of the board, and one
point for each committee (compensation, nominating, audit) that the
CEO sits on (source Board Analyst).
%BRD_EXPERT + % of independent directors that hold seats on other firms’ boards
(source Board Analyst).
%BRD_STOCK + % of directors that own stock in the firm (source Board Analyst).
GOVERNANCE_POLICY + One if the firm has a formal governance policy, zero otherwise (source
Board Analyst).
ARTICLE IN PRESS
Table 3 (continued )
Variables Predicted Definitions and data source

sign
%FINCOM_INSIDE + % of insiders on the finance committee (source proxy statement).

%NOM_IND + % of independent directors on the nominating committee (source
Board Analyst).
%COMP_IND + % of independent directors on the compensation committee (source
Board Analyst).
Firm characteristics
LEV Total debt (Compustat #9 plus Compustat #34) divided by total
assets (Compustat #6).
ROA + Net income before extraordinary items (Compustat #18) divided by
total assets.
LOSS One if the net income before extraordinary items is negative in the
current and prior fiscal year, zero otherwise.
INT_COV + Operating income before depreciation (Compustat #13) divided by
interest expense (Compustat #15) or (Compustat #339).
SIZE + Natural log of total assets.
SUBORD One if the firm has subordinated debt, zero otherwise.
CAP_INTEN + Gross PPE (Compustat #7) divided by total assets.
FIN_UTILITY + One if firm is a financial institution (one-digit SIC code 6) or a utility
(two-digit SIC code 49), zero otherwise.
Our first measure of financial transparency, WCAQ, is an accounting-based measure of

financial reporting quality based on the work of Dechow and Dichev (2002). Dechow and
Dichev (2002) derive a measure of working capital accruals quality that is a function of
how well working capital accruals map into cash flows. The calculation of WCAQ is a two-
step process beginning with the cross-sectional estimation of the following OLS model:
WCAt ¼ b0 þ b1 CFOt1 þ b2 CFOt þ b3 CFOtþ1 þ , (1)
where WCA is the working capital accruals ((Compustat # 302+Compustat #
303+Compustat # 304+Compustat # 305+Compustat # 307)) scaled by average total
assets (Compustat # 6), CFO is the cash flow from operations (Compustat # 308) scaled by
average total assets.
We estimate regressions by three, two, or one-digit SIC codes conditional on having at
least 10 firms in each SIC group. WCAQ is the standard deviation of the firm-specific
residual from the prior 3–5 years. Firms with smaller WCAQ values report working capital
accruals that better map into cash flows, i.e., are considered to have higher quality working
capital accruals. To facilitate discussion of our results, we multiply WCAQ by negative
one. Thus, larger WCAQ values imply higher quality working capital accruals.
Our second measure of financial transparency, TIMELINESS, is derived from the
following regression equation:
RETit ¼ b0 þ b1 NIBEit þ b2 LOSSit þ b3 NIBEit LOSSit þ b4 DNIBEit þ it , (2)
where RETit is the market-adjusted return for firm i over fiscal year t (from CRSP), NIBEit
is net income before extraordinary items (Compustat # 18) scaled by beginning of period
market value of equity for firm i in period t (Compustat # 25 * Compustat # 199), LOSSit
ARTICLE IN PRESS
is one if NIBE is negative, zero otherwise, DNIBEit is the change in net income before
extraordinary items scaled by beginning of period market value of equity, and
NIBEit*LOSSit is an interaction term that allows for a differential market reaction for
loss versus profit firms.
We estimate the above regression cross-sectionally within one-, two-, and three-digit SIC
codes requiring a minimum of 10 firms in each industry grouping.
Gu (2002) argues that the squared residuals from Eq. (2) can be conveniently interpreted
as the degree of price movement (returns) that is not explained by contemporaneous
accounting earnings. Higher squared residuals imply less timely earnings. To facilitate
interpretation of our results, we multiply this measure by negative one. Thus, larger (less
negative) values imply more timely earnings. One can think about financial statement
transparency as encompassing the relevance and reliability dimensions of accounting
information. The more information about the firm’s current economic activities that is
embedded in current earnings and the more precise that information (i.e., the more
relevant and reliable it is), then the more transparent the economic activities of a company
is to its stakeholders. Higher quality, more transparent earnings information means less
information asymmetry between the firm and its bondholders, leading to less uncertainty
about default risk which, in turn, should lead to higher credit ratings. Barth and Landsman
(2003) provide empirical support for this claim in that they find that firms with more value-
relevant earnings enjoy a lower cost of debt. We use the Gu measure as one proxy for
financial transparency as it captures both the timeliness of firms’ financial information and
relevance of the financial information for assessing firms’ current economic conditions.15
4.2.2. Control variables—firm characteristics

Additional firm-specific explanatory variables are included in the RATING models
based on a survey of prior research on the determinants of corporate bond ratings (e.g.,
Horrigan, 1966; Kaplan and Urwitz, 1979; Boardman and McEnally, 1981; Lamy and
Thompson, 1988; Ziebart and Reiter, 1992). The accounting-based ratios of debt-to-assets
(LEV), return-on-assets (ROA), and interest coverage (INT_COV) are used to proxy for
firms’ default risk, where higher LEV and lower ROA and INT_COV values reflect greater
default risk. We use LOSS, a categorical variable set equal to one if the firm reports
negative earnings in the current and prior fiscal year, as another control for default risk
given that the likelihood of default is greater for firms that are unprofitable.
Firm size, SIZE, is included as a control variable because larger firms face lower risk,
and thus are expected to have higher credit ratings. We also control for differences in firms’
debt structure by including SUBORD, which is coded one if the firm has subordinated
debt. The debt structure of a firm with subordinated debt is considered to more risky due
to the differential claims to assets by debt providers. Firms’ capital intensity
(CAP_INTEN) is included in the model to control for differences in firms’ asset structure,
where firms with greater capital intensity present lower risk to debt providers, and thus are
expected to have higher credit ratings. Finally, FIN_UTILITY is set equal to one if a firm
is a financial institution or utility firm, (zero for firms not in these two industries) to control
for lower default risk for firms operating in regulated industries.
15
To validate this construct, we regress the AIMR disclosure ratings on TIMELINESS for the same periods.
The coefficient is 13.26 with a t-statistic of 3.89. Thus, this measure of financial transparency and AIMR
disclosure rankings appears to be capturing similar constructs.
ARTICLE IN PRESS
The specific measurements of the variables capturing differences in firm characteristic

along with their predicted relation with RATING are summarized in Table 3.
4.2.3. Descriptive statistics

Table 4 presents descriptive statistics for the various governance and firm-characteristic
variables. Within the ‘‘Ownership Structure and Influence’’ component of governance, we
find the average (median) number of blockholders that own 5% or more or the firm’s stock
is 4.5 (4.0). The average (median) percentage of shares held by institutional investors is
63% (68%) while the average (median) percentage of shares held by insiders (officers and
Table 4
Summary statistics for credit rating variables
Variables Mean Standard Median 25% 75%

deviation

BLOCK 4.46 2.80 4.00 2.00 6.00
%INST 0.63 0.24 0.68 0.51 0.80
%INSIDE 0.08 0.11 0.04 0.02 0.08
Financial stakeholder rights and relations
G_SCORE 9.59 2.59 10.00 8.00 11.00
WCAQ 0.04 0.03 0.03 0.05 0.02
TIMELINESS 0.10 0.17 0.03 0.10 0.01
TOTFEES 0.11 0.22 0.04 0.02 0.10
%AUD_IND 0.92 0.16 1.00 1.00 1.00
FIN_EXPERT 0.26 0.44 0.00 0.00 1.00
%BRD_IND 0.70 0.16 0.73 0.58 0.83
CEOPOWER 0.78 0.50 1.00 0.00 1.00
%BRD_EXPERT 0.36 0.22 0.36 0.20 0.50
%BRD_STOCK 0.87 0.20 0.92 0.82 1.00
GOVERNANCE_POLICY 0.42 0.49 0.00 0.00 1.00
%FINCOM_INSIDE 0.15 0.12 0.13 0.08 0.22
%NOM_IND 0.79 0.32 1.00 0.67 1.00
%COMP_IND 0.90 0.20 1.00 0.83 1.00
LEV 0.31 0.17 0.30 0.19 0.41
ROA 0.02 0.08 0.03 0.01 0.06
LOSS 0.13 0.34 0.00 0.00 0.00
INT_COV 10.78 18.77 5.16 2.79 10.23
SIZE 8.46 1.49 8.20 7.38 9.44
SUBORD 0.18 0.39 0.00 0.00 0.00
CAP_INTEN 0.54 0.40 0.48 0.19 0.84
FIN_UTILITY 0.23 0.42 0.00 0.00 0.00
RATING 3.83 1.11 4.00 3.00 5.00
INVESTMENT_GRADE 0.63 0.48 1.00 0.00 1.00
Notes: RATING ¼ S&P LT Domestic Issuer Credit Rating (Compustat #280), see Table 1 for numeric coding.
INVESTMENT_GRADE ¼ 1 if a firm’s credit rating is investment grade as noted in Table 1, zero otherwise. See
Table 3 for other variable definitions.
ARTICLE IN PRESS
directors) is 8% (4%). For the ‘‘Financial Stakeholder Rights’’ dimension of corporate

governance, the average G_SCORE of our sample firms is 9.59, which is similar to the
mean G_SCORE reported by Gompers et al. (2003) of 9.15. Sixty-one of our sample firms
(6.8%) fall into Gompers et al.’s dictatorship portfolio (G_SCORES413 indicating
greater management power) while 41 firms (4.7%) fall into their democracy portfolio
(G_SCORESo6 indicating greater shareholder rights).
Turning to the ‘‘Financial Transparency’’ dimension, the mean (median) WCAQ and
TIMELINESS value is 0.04 (0.03) and 0.10 (0.03), respectively. Recall the larger
(i.e., less negative) values of these variables reflect higher quality financial reporting. The
measure of economic bonding between the firm and its auditor is our TOTFEES
variable.16 Because of its small magnitude, we multiply this variable by 100. Before this
scaling adjustment, the median firm’s total fees paid to its auditor amount to only 0.04%
of the audit firm’s total revenues. Ninety-two percent of the average firm’s audit committee
is comprised of outside independent board members, with over three-quarters of the
sample firms having 100% independent audit committees. The little variation in
%AUD_IND is explained by the fact that the listing rules of the New York Stock
Exchange and NASDAQ require listed firms to have audit committees comprised of at
least three independent directors. There are, however, exceptions to the rules that result in
not all firms having 100% independent audit committees (see Klein (2003) for an
overview). Finally, 26% of our sample firms have an outside financial expert (CPA or
CFO) serving on their audit committee.
Within the ‘‘Board Structure and Processes’’ dimension, the descriptive statistics
indicate that the average (median) percentage of outsiders on the board is 70% (73%) and
the lower quartile value is 58%. Consistent with the evidence in Table 2, the majority of
our sample firms have CEOs that also serve as Chairman of the Board or on other board
committees. On average, 36% of outside directors serve on other boards and 87% of the
directors hold stock in the company. Forty-two percent of the sample firms have a formal
governance policy. The average percentage of outsiders on the compensation (nominating)
committee is 90% (79%), while the average percentage of insiders on the finance
committee is 15%.17
For brevity, we do not describe the summary statistics for the firm-characteristic
variables. Turning to the dependent variables, we note that the median credit rating is 4.0
implying a debt rating in the BBB+ to BBB range, and that 63% of our sample firms
have an investment-grade credit rating.
Table 5 presents correlations among the firm-characteristic variables (Panel A) and
governance variables (Panel B) and between these variables and credit ratings. The upper
right-hand portion of each panel presents Pearson product–moment correlations while the
lower left-hand portion presents the Spearman rank-order correlations. In Panel A, the
simple correlations between each of the firm characteristics and our RATING variable are
in the predicted directions and are statistically significant at the 0.01 level or below except
for the capital intensity variable (CAP_INTEN) which is negative and insignificant.
16
Recall we measure this as total fees (audit and non-audit) paid by the client divided by the audit firm’s total
revenue.
17
Recall that in coding this variable, if a firm does not have a standing finance committee, we used the
percentage of insiders on the overall board for %FINCOM_INSIDE because the board de facto votes on all
major financing decisions in the absence of a finance committee. This explains why the percentage of insiders on
this committee appears to be smaller than the numbers imply in Table 2.
Table 5
Correlations
Panel A: Firm characteristics

RATING LEV ROA LOSS INT_COV SIZE SUBORD CAP_INTEN FIN_UTILITY
RATING 0.28 0.44 0.42 0.27 0.50 0.19 0.03 0.24

LEV 0.26 0.22 0.21 0.42 0.03 0.21 0.28 0.03
ROA 0.39 0.23 0.60 0.34 0.06 0.04 0.01 0.00
LOSS 0.41 0.17 0.55 0.18 0.16 0.01 0.06 0.18
INT_COV 0.45 0.52 0.72 0.42 0.02 0.14 0.07 0.13
SIZE 0.51 0.06 0.09 0.15 0.04 0.02 0.20 0.41
SUBORD 0.21 0.19 0.13 0.01 0.19 0.06 0.09 0.04
CAP_INTEN 0.04 0.33 0.01 0.08 0.03 0.19 0.10 0.18
FIN_UTILITY 0.26 0.03 0.16 0.18 0.23 0.39 0.04 0.23
Panel B: Governance variables

A B C D E F G H I J K L M N O P Q R
RATING A 0.36 0.11 0.16 0.19 0.30 0.38 0.26 0.03 0.06 0.18 0.16 0.11 0.07 0.31 0.25 0.26 0.10
BLOCK B 0.39 0.22 0.14 0.05 0.18 0.09 0.20 0.01 0.02 0.05 0.06 0.01 0.06 0.06 0.12 0.10 0.04
%INST C 0.01 0.35 0.18 0.13 0.08 0.07 0.02 0.06 0.08 0.18 0.21 0.12 0.02 0.19 0.04 0.14 0.04
%INSIDE D 0.33 0.29 0.08 0.23 0.10 0.04 0.15 0.04 0.12 0.39 0.31 0.21 0.06 0.24 0.13 0.15 0.19
G_SCORE E 0.21 0.06 0.09 0.18 0.10 0.14 0.07 0.09 0.09 0.27 0.25 0.18 0.05 0.20 0.23 0.12 0.20
WCAQ F 0.32 0.23 0.18 0.22 0.07 0.16 0.09 0.01 0.07 0.08 0.02 0.07 0.02 0.07 0.17 0.07 0.04
ARTICLE IN PRESS
TIMELINESS G 0.28 0.11 0.02 0.06 0.06 0.16 0.02 0.04 0.03 0.02 0.02 0.07 0.08 0.04 0.13 0.10 0.01
TOTFEES H 0.25 0.15 0.05 0.29 0.04 0.03 0.01 0.06 0.06 0.09 0.07 0.05 0.07 0.28 0.06 0.20 0.05
FIN_EXPERT I 0.04 0.01 0.05 0.05 0.10 0.01 0.04 0.02 0.09 0.10 0.14 0.09 0.01 0.05 0.01 0.04 0.05
%AUD_IND J 0.03 0.04 0.08 0.11 0.07 0.00 0.04 0.05 0.09 0.51 0.38 0.41 0.04 0.27 0.06 0.09 0.12
%BRD_IND K 0.19 0.04 0.15 0.38 0.27 0.06 0.05 0.13 0.10 0.46 0.56 0.55 0.13 0.54 0.16 0.23 0.49
%NOM_IND L 0.14 0.04 0.18 0.23 0.20 0.04 0.01 0.06 0.13 0.42 0.55 0.44 0.02 0.36 0.17 0.24 0.26
%COMP_IND M 0.10 0.00 0.10 0.18 0.16 0.03 0.06 0.06 0.10 0.43 0.51 0.50 0.00 0.31 0.09 0.14 0.14
CEOPOWER N 0.07 0.06 0.01 0.12 0.06 0.02 0.06 0.09 0.01 0.05 0.16 0.01 0.02 0.12 0.07 0.06 0.09
%BRD_EXPERT 0 0.31 0.05 0.17 0.35 0.21 0.01 0.03 0.35 0.05 0.27 0.54 0.36 0.31 0.13 0.14 0.33 0.33
H. Ashbaugh-Skaife et al. / Journal of Accounting and Economics 42 (2006) 203–243
%BRD_STOCK P 0.28 0.16 0.02 0.19 0.28 0.19 0.05 0.13 0.01 0.08 0.24 0.19 0.15 0.10 0.22 0.15 0.09
GOVERNANCE POLICY Q 0.25 0.10 0.11 0.21 0.12 0.05 0.07 0.21 0.04 0.08 0.22 0.21 0.14 0.08 0.33 0.18 0.23
%FINCOM_INSIDE R 0.15 0.05 0.01 0.24 0.22 0.06 0.02 0.16 0.06 0.11 0.49 0.18 0.14 0.07 0.35 0.18 0.23
Notes: Bold text indicates significance at the 0.01 level or better. RATING ¼ S&P LT Domestic Issuer Credit Rating (Compustat #280), see Table 1 for numeric
223
coding. See Table 3 for other variable definitions.

ARTICLE IN PRESS
Specifically, we find that ROA, INT_COV, SIZE, and FIN_UTILITY are significantly
positively correlated with credit ratings whereas LEV, LOSS, and SUBORD are
significantly negatively correlated with ratings. Not surprisingly, several of the firm-
characteristic variables exhibit high intercorrelations.
Panel B of Table 5 presents the correlations between the various governance variables
and between these variables and RATING. Fourteen of the 16 governance variables
exhibit Pearson correlations with the RATING variable that are significant at 0.01 or
better. The correlations among the various governance variables generally fall below 0.30
except for the board and committee independence measures (shown in shaded cells) which
are generally in the 0.38–0.56 range. The high intercorrelations between the committee and
board independence measures are to be expected because the committees are drawn from
the board membership. Because of these high correlations, we include only the board and
audit committee independence measures in our logit model.
5. Empirical tests and results
5.1. Ordered logit results
Our empirical tests are derived from a general model that represents credit ratings as a
function of firm characteristics and corporate governance attributes.
Credit rating ¼ f ðfirm characteristics; corporate governance attributesÞ.
To test the predicted relations between corporate governance attributes and credit ratings,
we estimate an ordered logit model. We use ordered logit because the seven categories of
credit ratings convey ordinal risk assessments; we can rank order firms’ preferences across
the rating categories but cannot assume uniform differences in benefits (costs) between the
categories.
The first column of Table 6 reports the full model, where we jointly test whether firm risk
characteristics and governance attributes that capture the four components of the S&P
corporate governance framework are associated with firms’ credit ratings. The model is
highly significant with a likelihood ratio w2 of 811.55 and a generalized R2 of 0.60. The
coefficients on the firm characteristics have the expected sign and, except for
FIN_UTILITY, are significant at the 0.01 level or better. Consistent with prior research
(Horrigan, 1966; Kaplan and Urwitz, 1979), we find that credit ratings are positively
related to ROA, INT_COV, SIZE, CAP_INTEN, and negatively related to LEV, LOSS,
and SUBORD.
Turning to the governance attributes, the highly significant Wald w2 indicates that as a
group the governance variables add significant explanatory power over and above the firm-
level risk characteristics. Further, within each component of the S&P governance
framework, there is at least one dominant governance attribute that affects firms’ credit
ratings. Specifically, when simultaneously considering all attributes representing the four
components of governance, we find BLOCK, G_SCORE, WCAQ, TIMELINESS,
%BRD_IND, CEOPOWER, %BRD_EXP, and %BRD_STOCK to be significant
determinants of credit ratings.
The negative coefficient on BLOCK indicates that firms with a larger number of
blockholders have lower credit ratings. This finding is consistent with our ‘‘wealth
redistribution’’ hypothesis that blockholders, as influential shareholders, can exercise
ARTICLE IN PRESS
Table 6
Logistic regression results of the effects of corporate governance attributes on firms’ credit ratings (dependent
variable ¼ RATING)
Variables Predicted Estimated coefficient
Sign Model 1 Model 2 Model 3
LEV 2.014*** 2.223***
ROA + 0.094*** 0.108***
LOSS 0.986*** 1.073***
INT_COV + 0.017*** 0.015***
SIZE + 0.643*** 0.869***
SUBORD 0.604*** 0.744***
CAP_INTEN + 0.538*** 0.863***
FIN_UTILITY + 0.214 0.436***
Governance attributes
BLOCK ? 0.165*** 0.243***
%INST ? 0.390 1.209***
%INSIDE 0.442 0.603
G_SCORE ? 0.037* 0.050*
Financial transparency and information disclosure

WCAQ + 7.372*** 14.301***
TIMELINESS + 3.463*** 4.502***
TOTFEES 0.340 1.576***
%AUD_IND + 0.036 0.618
FIN_EXPERT + 0.126 0.157
%BRD_IND + 0.950* 0.307
CEOPOWER 0.177* 0.098
%BRD_EXPERT + 1.003*** 1.930***
%BRD_STOCK + 1.109*** 1.298***
GOVERNANCE_POLICY + 0.179 0.302**
%FINCOM_INSIDE + 0.317 0.252
Generalized R2 0.60 0.52 0.41
Likelihood ratio w2 811.55*** 660.55*** 474.32***
Wald w2 139.93***
Sample size 894 894 894
Notes: Table values are coefficient estimates from the following ordered logit models:
Model 1: RATING ¼ f (firm characteristics, governance attributes).
Model 2: RATING ¼ f (firm characteristics).
Model 3: RATING ¼ f (governance attributes).
***,**,* indicate significance at the 0.01, 0.05, and 0.10 level or better, respectively. The Wald w2 statistic tests
whether the governance attributes added in each model, as a whole, explain a significant portion of the variation
in firms’ credit ratings. RATING ¼ S&P LT Domestic Issuer Credit Rating (Compustat #280), see Table 1 for
numeric coding. See Table 3 for other variable definitions.
ARTICLE IN PRESS
influence over management to secure benefits that are detrimental to bondholders. This
result is also consistent with the findings of Bhojraj and Sengupta (2003) who document
that blockholders have an adverse impact on bond ratings.
We find a positive and marginally significant coefficient on G_SCORE. Recall that the
smaller the G_SCORE, the stronger the shareholder rights. Our results suggest that
stronger shareholder rights are associated with lower firm credit ratings. Gompers et al.
(2003) find that firms with stronger shareholder rights have higher firm value and higher
profits. Thus, our finding is most consistent with the ‘‘wealth redistribution’’ hypothesis
that posits that bondholders may suffer potential wealth transfer effects associated with
stronger shareholder rights. That is, from the bondholders’ perspective, the risks of wealth
transfer that can result with stronger shareholder rights outweigh the positive firm value
effects documented in Gompers et al. (2003). Our finding of a positive association between
G_SCORE and credit ratings is consistent with the work of Asquith and Wizman (1990)
and Warga and Welch (1993) who find that ownership changes can result in significant
wealth transfers from bondholders to shareholders. Our results are also consistent with
Klock et al. (2004) who find that firms with stronger anti-takeover provisions (weaker
shareholder rights) have a lower cost of debt financing relative to firms with weaker anti-
takeover provisions.
As predicted, we find a positive relation between RATING and the transparency of
firms’ financial reporting as measured by WCAQ and TIMELINESS. However, we find no
evidence that the quality of the audit process affects a firm’s credit rating. Specifically, we
do not find that firms having more independent directors serving on their audit committees
and having an independent financial expert on the audit committee have better credit
ratings. Moreover, we do not find a significant association between TOTFEES and
RATING.
In general, the estimated coefficients on the ‘‘Board Structure and Processes’’ variables
support our ‘‘management disciplining’’ hypothesis that better board structure and
processes contribute to more efficient monitoring of management that protects all
stakeholders’ interests. As predicted, we find a positive coefficient on %BRD_IND, which
indicates that the greater the board’s ability to provide independent oversight of
management, the better the credit rating. This result is consistent with Bhojraj and
Sengupta (2003) who find that firms with a greater proportion of independent outside
directors on the board have higher bond ratings. We also find a marginally significant
negative coefficient on CEOPOWER. This result suggests that it is costly for firms, in
terms of default risk, to cede the chief executive officer with too much board control. The
positive coefficient on %BRD_EXPERT indicates that when a greater proportion of the
board is comprised of knowledgeable individuals, as proxied by their service to other
boards, the higher the firm credit rating. We also document a positive relation between
%BRD_STOCK and RATING. This result indicates that as more members of the board
have an equity stake in the firm, they have greater incentives to restrict managerial
opportunism or to monitor management decision making leading to lower default risk.
To benchmark our results against prior credit rating studies, the Model 2 column of
Table 6 reports results for the firm risk characteristics in isolation. Omitting the
governance variables reduces the generalized R2 in Model 1 from 0.60 to 0.52 and lowers
the likelihood ratio w2 from 811.55 to 660.55. Model 3 column of Table 6 reports results
using only the governance variables to explain credit ratings. Interestingly, the governance
variables explain nearly as much of the variation in credit ratings as do the firm risk
ARTICLE IN PRESS
characteristics (generalized R2 of 0.41 versus 0.52). In summary, we find that governance

attributes explain a significant proportion of the cross-sectional variation in credit ratings.
Moreover, the explanatory power of governance attributes, as a group, is comparable to
firm risk characteristics that have garnered most of the attention in prior research.
5.2. Investment versus speculative-grade analysis
As stated above, credit ratings convey ordinal risk assessments. Because of the difficulty
in quantifying the marginal effects of changes in each governance variable on credit ratings
with multiple categories, we use an alternative classification scheme that partitions credit
ratings into two categories—investment grade or speculative grade. Many bond portfolio
managers are restricted from owning speculative-grade bonds (Grinblatt and Titman,
2002), and as such, firms incur significant costs if they receive a speculative bond rating.
Furthermore, using a dichotomous credit rating classification allows us to more readily
assess the economic impact of corporate governance on firms’ expected cost of debt.
Table 7 displays the results of estimating the logitistic regression using INVEST-
MENT_GRADE as the dependent variable, where INVESTMENT_GRADE is coded
one if the firm’s credit rating is BBB or better, and zero otherwise. The results are similar to
the results of the RATING analyses reported in Table 6 in that we find that BLOCK,
G_SCORE, WCAQ, TIMELINESS, and %BRD_STOCK are significant determinants of
INVESTMENT GRADE. In contrast, the coefficient on INT_COV is insignificant in the
INVESTMENT_GRADE analysis whereas it was highly significant in the RATING
analyses. In addition, the coefficients on %BRD_EXPERT, %BRD_IND, and CEO-
POWER are no longer significant when using INVESTMENT_GRADE as the proxy for
credit risk.
In order to provide some insight into the economic significance of our results, we
calculate the change in probability of receiving an investment-grade credit rating as a result
of changing the levels of various corporate governance variables. The change in probability
is calculated using the following steps. First, we calculate the probability of achieving an
investment-grade credit rating from our logitistic regression model using the following
expression:
0
. 0
pðX Þ ¼ eb X ð1 þ eb X Þ, (3)
where b is the vector of coefficients reported in Table 7 and X is the vector of independent
variables set equal to their mean values. Next, we calculate the marginal changes in the
probability of a firm receiving an investment-grade credit rating as a result of a one
standardized unit change in each of our governance variables. This marginal effect is
measured by qpðX Þ=qxi ¼ bi pðX Þ½1 pðX Þ, which is calculated at the mean value of the
regressors. These marginal effects are reported in column 3 of Table 8 for the governance
variables after standardizing each non-binary variable by its mean and dividing by its
standard deviation.18 The marginal effects measure the change in the probability of
receiving an investment-grade rating for a one standardized unit change in each
governance variable while holding the firm characteristics at their mean values.
18
We use standardized values because the various governance variables are measured in different units. Without
standardization, the marginal probabilities are difficult to compare and interpret (Agresti 2002, Chapter 5).
ARTICLE IN PRESS
Table 7
Logistic regression results of the effects of corporate governance attributes on firms’ credit ratings (dependent
variable ¼ INVESTMENT_GRADE)
Variables Predicted sign Estimated coefficient
INTERCEPT ? 7.125***
LEV 3.545***
ROA + 0.130***
LOSS 1.111***
INT_COV + 0.007
SIZE + 0.913***
SUBORD 1.416***
CAP_INTEN + 0.925***
FIN_UTILITY + 0.194
BLOCK ? 0.226***
%INST ? 0.583
%INSIDE 0.207
G_SCORE ? 0.067*
Financial transparency and information disclosure
WCAQ + 6.995**
TIMELINESS + 4.108***
TOTFEES 0.110
%AUD_IND + 0.434
FIN_EXPERT + 0.004
%BRD_IND + 0.885
CEOPOWER 0.168
%BRD_EXPERT + 0.698
%BRD_STOCK + 1.320***
GOVERNANCE_POLICY + 0.066
%FINCOM_INSIDE + 0.224
Generalized R2 0.48
Likelihood ratio w2 587.35***
Wald w2 72.84***
Sample size 894
whether the governance attributes added in each model as a whole explain a significant portion of the variation in
firms’ investment-grade ratings. INVESTMENT_GRADE ¼ 1 if a firm’s credit rating is investment grade as
noted in Table 1, zero otherwise. See Table 3 for other variable definitions.
An alternative way of assessing the effect of various governance variables on the

likelihood of receiving an investment-grade credit rating that is easier to interpret is to
calculate the values of the logit function, pðX Þ, at selected xi values such as their lower and
upper quartiles (Agresti, 2002, p. 167). This entails substituting the quartile values for each
ARTICLE IN PRESS
Table 8
Assessment of changes in probabilities of receiving an investment-grade credit rating
Variables Predicted sign Marginal effect Change in probability

standardized Q1 versus Q3 values
variables
BLOCK ? 0.136 0.194
%INST ? 0.030 0.036
%INSIDE 0.005 0.003
G_SCORE ? 0.037 0.043
WCAQ + 0.042 0.045
TIMELINESS + 0.152 0.081
TOTFEES 0.005 0.002
%AUD_IND + 0.015 0.000
FIN_EXPERT + 0.001 0.001
%BRD_IND + 0.031 0.048
CEOPOWER 0.018 0.036
%BRD_EXPERT + 0.033 0.045
%BRD_STOCK + 0.056 0.052
GOVERNANCE_POLICY + 0.014 0.014
%FINCOM_INSIDE + 0.006 0.007
Notes: The Marginal effects column shows the effects of receiving an investment-grade credit rating due to a one
unit change in the variable
. of interest after standardizing the independent variables. Marginal effects are
0 0
computed as: pðX Þ ¼ eb X ð1 þ eb X Þ where b0 X is evaluated at the mean values of X. Tabled values in the Change
in probability column show the change in the probability of receiving an investment-grade credit rating as a result
of moving from the first to the third quartile value of the variable of interest, holding all other variables constant
at their mean values. See Table 3 for variable definitions.
xi explanatory variable into Eq. (2) while holding the other variables constant at their
means. The linear approximation to changes in pðX Þ is obtained by multiplying the
interquartile range of xi values (see Table 4 for the interquartile ranges) by the marginal
effects based on the unstandardized value of the variables (Agresti, 2002, Chapter 5). These
values are reported in the last column of Table 8.
Moving from the first quartile to the third quartile of BLOCK decreases the probability
of receiving an investment-grade credit rating by approximately 0.19. The changes in
probabilities for G_SCORE, WCAQ, and TIMELINESS are approximately 0.04, 0.05,
and 0.08, respectively, while the changes in probabilities for %BRD_IND and
%BRD_STOCK are 0.048 and 0.052, respectively. Although the probability changes
due to any one governance variable may not appear to be all that dramatic, the aggregate
effect across all dimensions of corporate governance can be substantial.
To demonstrate this point, we first calculate the probability of receiving an investment-
grade credit rating for a hypothetical firm that takes on the lower (upper) quartile values of
ARTICLE IN PRESS
governance variables that are positively (negatively) related to credit ratings while holding
all the firm-specific variables at their mean values.19 This yields a probability of receiving
an investment-grade credit rating of 0.46. We next repeat this process but now use upper
(lower) quartile values of governance variables that are positively (negatively) related to
credit ratings. This yields a probability of receiving an investment-grade credit rating of
0.93. Thus, a firm approximately doubles its probability of receiving an investment-grade
credit rating by implementing better governance along multiple dimensions.20 During the
time frame of our analysis, the average yield for firms with investment (speculative)-grade
debt with a 10-year maturity was approximately 6.00% (14.0%). This 800-basis point
spread translates into $38.4 million in before-tax interest costs for the median firm in our
investment-grade sample.21 This cost estimate, however, overstates the cost of weak
governance from shareholders’ perspective when weak governance allows wealth transfers
from bondholders to shareholders to take place. To the extent that part of the interest rate
spread between investment-grade and speculative-grade debt is due to bondholders price-
protecting against potential wealth transfers between themselves and stockholders, the
amount computed above represents an upper bound on the cost of weak governance from
the stockholder’s perspective. Nevertheless, governance mechanisms that increase firms’
likelihood of receiving an investment-grade debt rating have significant implications for
assessing debt financing costs.
6. Additional analyses
6.1. Why don’t all firms practice good governance?
The preceding analysis suggests that firms with weak governance have a lower
probability of receiving an investment-grade credit rating and, as a consequence, incur
significantly higher debt costs. This raises the question of why some firms are willing to
bear additional debt financing costs by not practicing good governance. One way to think
about answering this question is to consider how managers can appropriate some or all of
the rents from outside stakeholders by resisting better governance.
Evidence by Core et al. (1999) suggests one way that managers can extract rents from
weak governance. They find that CEOs with greater power over the board or that are more
entrenched earn greater compensation after controlling for standard economic determi-
nants of pay. Moreover, they find that the estimated component of overcompensation is
significantly negatively related with subsequent firm operating and stock performance.
Their results suggest that firms with weaker governance structures exhibit greater
19
For governance attributes measured as a 0–1 dummy variable, the benchmark probability is determined with
the zero (one) value when the governance attribute is negatively (positively) related to credit ratings.
20
We hasten to note that this illustration does not reflect the typical firm in our sample because any given firm
will likely not start from a position of having weak governance (low quartile) along all of the multiple dimensions
we consider. Nor is it likely that any given firm would be able to move to a position of having strong governance
along all dimensions (upper quartile).
21
To calculate the interest savings, we first determine the mean debt-to-asset ratio for our sample of investment-
grade firms (0.28). We then multiply this ratio times the total assets of each speculative-grade firm to estimate the
‘‘as-if’’ capital structure if the speculative-grade firm was an investment-grade firm. Finally, we multiply the
estimated debt level times the 800-basis point spread to determine a speculative firm’s hypothetical interest
savings.
ARTICLE IN PRESS
overcompensation of CEOs and face greater agency problems. Generalizing the Core et al.
(1999) results to our research setting, it may be rational for managers to resist efforts to
improve governance as long as they receive more overcompensation relative to their share
of increased debt costs due to weaker governance. Moreover, various frictions in the
market for corporate control such as anti-takeover amendments, super-majority voting
rules for ownership changes, poison pills, and golden parachutes contribute to manage-
ment entrenchment and allow overcompensation and non-optimal governance structures
to persist.
To investigate whether overcompensation helps explain why some firms do not seek
stronger governance, we model CEO pay in 2002 as
CEO_PAY ¼ f ðeconomic determinants; board and ownership structure attributesÞ.
Following Core et al. (1999), we measure CEO_PAY in three different ways: Salary,
Salary+Bonus and Total Compensation. The definitions of the alternative measures of
CEO_PAY, the specific board and ownership structure variables and the economic
determinants are detailed in the Appendix. As noted in Core et al. (1999), the portion of
CEO_PAY explained by the board and ownership variables represents overcompensation.
Under the optimal contracting view, there should be no association between these
governance provisions and CEO compensation, i.e., CEO compensation is only a function
of economic determinants.
The results of estimating the cross-sectional models of CEO_PAY for sample firms
(Eq. (A1)) are presented in Panel A of Table 9. As shown, the economic and board and
ownership variables explain from 39% (for Total Compensation) to 52% (for
Salary+Bonus) of the variation in CEO pay. The F-tests on the incremental explanatory
power of the set of board and ownership variables relative to the economic determinants of
pay are significant at 0.001 level or better.
To estimate the proportion of the CEO’s compensation that represents overcompensa-
tion (OVERCOMP), we first calculate for each sample firm the predicted excess
compensation by multiplying the estimated board and ownership coefficients by the
sample firm’s board and ownership variables’ values. We then scale predicted over-
compensation (OC) by the relevant CEO_PAY value (which we label OC_Salary,
OC_Salary+Bonus and OC_TotalComp).
Our major findings related to overcompensation are summarized in Panel B of Table 9.
The mean (median) overcompensation percentage for firms in our Investment-Grade
sample ranges from 34.9% (32.7%) for OC_Salary to 56.1% (50.9%) for OC_TotalComp.
For the firms in our Speculative-Grade sample, the corresponding mean (median)
overcompensation percentages range from 39.1% (35.5%) for OC_Salary to 59.2%
(60.8%) for OC_TotalComp. For two of the three compensation measures, the
Speculative-Grade sample exhibits significantly greater overcompensation of CEOs
relative to the Investment-Grade sample after controlling for standard economic
determinants of pay.
To give some idea of the incentives that CEOs of Speculative-Grade firms have for
trading off higher firm debt cost due to weaker governance for higher overcompensation
paid directly to them, we scale the CEO’s overcompensation by his share of additional
financing costs that the firm incurs by having Speculative-Grade debt. To calculate the
CEO’s share of the higher financing costs, we multiply the CEO’s percentage ownership in
ARTICLE IN PRESS
Table 9
Governance and CEO compensation
Panel A: OLS model of CEO compensation

Dependent variable
Salary Salary+Bonus Total compensation
Economic determinants
SALES 123.143*** 331.884*** 1873.306***
MB 1.589 13.675 104.749*
OPROA 223.394 813.621 774.941
RET 29.145 199.105*** 287.823
STD_ OPROA 57.652 251.676 3489.100
STD_RET 43.506 54.487 1130.478**
Governance determinants
CEOPOWER 51.177*** 103.168 583.542*
%COMP_CEOAPP 78.960*** 485.927*** 905.928**
BRDSIZE 13.835*** 45.726*** 146.444**
%BRD_INSIDE 98.787 248.887 853.844
%OUT_BUSY 116.613 590.452** 1552.127
%BRD_ActiveCEOs 43.982 64.081 684.285
CEO Ownership 244.240** 190.693*** 22.806
%INST 1.121** 3.832** 11.731
BLOCK 4.229 47.030*** 202.682***
Adj R2 0.47 0.52 0.39
F-stat governance 5.65*** 8.21*** 2.73***
determinants
Panel B: Estimates of overcompensation by grade

Overcompensation estimates INVESTMENT_GRADE SPECULATIVE_GRADE P-value
n ¼ 492 n ¼ 245 (two-
tailed)
OC_Salary % (mean) 0.349 0.391 0.00

OC_Salary % (median) 0.327 0.355 0.01
OC Salary+Bonus %(mean) 0.535 0.600 0.00
OC Salary+Bonus %(median) 0.478 0.573 0.01
OC_Total Comp %(mean) 0.561 0.592 0.24
OC_Total Comp %(median) 0.509 0.608 0.26
Panel C: Overcompensation to share of interest costs ratio—speculative grade firms (n ¼ 245)

Q1 Median Q3
OC_Salary/CEOSHARE_INTEXP 51.242 122.816 435.501

OC Salary+Bonus /CEOSHARE_INTEXP 133.467 334.230 1212.920
OC_Total Comp /CEOSHARE_INTEXP 365.773 905.243 3180.150
ARTICLE IN PRESS
Table 9 (continued)
Panel D: Test of significance
No. of firms No. of firms w2
where ratio is where ratio is
less than or greater than 1
equal to 1
OC_Salary/CEOSHARE_INTEXP 5 240 225.41***

OC Salary+Bonus /CEOSHARE_INTEXP 4 241 229.26***
OC_Total Comp /CEOSHARE_INTEXP 4 241 229.26***
Notes: Panel A presents the coefficient estimates for OLS regression, see Appendix. ***,**,* indicate significance
at the 0.01, 0.05, and 0.10 level or better, respectively. Salary is equal to the dollar value in thousands of the base
salary (cash and non-cash) earned by the CEO during fiscal 2002. Salary+Bonus is equal to the current
compensation in thousands comprised of salary and bonus earned by the CEO during fiscal 2002. Total
Compensation is equal to total compensation in thousands earned by the CEO during fiscal 2002, which is
comprised of the following: salary, bonus, other annual pay, total value of restricted stock granted, total value of
stock options granted (using Black-Scholes), long-term incentive payouts, and all other total pay. OC_Salary % is
defined as the dollar amount of Salary due to the governance determinants divided by salary; Salary+Bonus % is
defined as the dollar amount of Salary+Bonus due to the governance determinants divided by Salary+Bonus;
OC_Total Comp % is defined as the dollar amount of Total Compensation due to the governance determinants
divided by Total Compensation. INVESTMENT_GRADE and SPECULATIVE_GRADE are based on a firm’s
credit rating as noted in Table 1. Median differences are assessed using the Wilcoxon rank sum test for differences
in the distributions. All P-values are two sided. To calculate the CEO’s portion of avoidable interest
(CEOSHARE_INTEXP), we multiply 8% (spread between investment-grade and non-investment-grade debt)
times 0.65 (tax benefit to debt) times the CEO’s ownership stake in the firm times the total debt outstanding. For
CEOs with zero ownership in the firm, we set the OC/CEOSHARE_INTEXP to the sample median for the
respective compensation figure. See Table 3 for other variable definitions.
the firm by the after-tax additional financing costs to the firm.22 The results of this
comparison are presented in Panel C of Table 9. The median value of OC_Salary is
$240,000 (untabulated) which is roughly 123 times the CEO’s share of the additional after-
tax interest cost from having speculative-grade debt versus investment-grade debt. For
OC_Salary+Bonus, the median overcompensation is $646,000 with a multiple of 334, and
for OC_TotalComp the median overcompensation is $1,808,000, which translates into a
multiple 905 times the CEO’s share of the after-tax interest cost differential.23
To statistically test the CEO’s benefits-to-cost ratio of weak governance, we partition
firms in our Speculative-Grade sample into two groups. The first group represents the sub-
sample of firms for which the CEO’s overcompensation is less than or equal to the CEO’s
share of the interest expense. The other group is the sub-sample of firms for which the
CEO’s overcompensation is greater than his portion of interest costs due to weak
governance. The latter group reflects the firms for which the CEO benefits outweigh the
costs of weak governance. Consequently, these are firms where the CEO is expected to
22
The additional after-tax financing costs are determined by multiplying the firm’s long-term debt at the
beginning of 2002 times 8% (the spread between investment-grade and speculative-grade debt with 10 year
maturity in 2002) times 65% (1marginal corporate tax rate).
23
A major reason for these relatively large multiples is because CEOs of most firms in the speculative-grade
sample hold such a small percentage of the firm’s shares. For example, 169 of the 245 speculative-grade firms have
CEOs that own less than one-tenth of 1% of the firm’s stock. Thus, the typical CEO’s share of the higher after-tax
interest costs is quite small.
ARTICLE IN PRESS
impede governance improvements as they would result in a net loss to the CEO, given his
degree of overcompensation and ownership stake in the firm. The w2 test results suggest
that for the vast majority of firms in the Speculative-Grade debt sample, the degree of CEO
overcompensation outweighs his share of the additional debt costs that may result from
weaker governance.
A competing explanation for why CEOs may be willing to have the firm incur higher
interest costs due to weaker governance is that the higher interest costs are associated with
the firm taking on riskier projects with a higher expected return, which benefits
shareholders as the residual claimants. The higher interest costs could be offset by higher
shareholder value, which can also benefit the CEO indirectly via continued employment or
directly via his personal ownership stake in the firm. Moreover, under this competing
explanation the ‘‘excess compensation’’ that we have estimated may simply be
compensating managers for bearing higher risk.
If this risk explanation for higher compensation is valid, we should observe firms with
lower credit ratings exhibiting higher future performance. That is, we would expect riskier
firms to earn higher accounting rates of return. Table 10 provides evidence on this
competing explanation for our sample of firms. Panel A shows that next year’s ROA
(ROE) is positively associated with firms having investment-grade credit ratings after
controlling for the current year ROA (ROE). Thus, it is not the case that firms with lower
credit ratings have higher shareholder value due to stronger future performance.
Consistent with results in Core et al. (1999), Panel B shows a significant negative relation
between the amount of overcompensation and future period ROA. Moreover, our excess
compensation results are not due to risk premia—that is, compensating managers for
bearing higher risk. The risk effects on compensation are controlled for through two of the
economic determinants of compensation included in the model–the standard deviation of
ROA (STD_OPROA) and the standard deviation of stock returns (STD_RET). These
results are consistent with the conjecture that firms with weaker governance have greater
agency problems and the resulting CEO overcompensation is a dead weight cost that is
borne by all stakeholders.
If it is easier for CEOs of firms with weaker governance to garner excess compensation,
as the results presented here suggest, and if the CEO’s share of the additional debt costs are
low, then there are clear incentives for managers to resist efforts to strengthen governance.
Thus, CEO expropriation of rents through excess compensation provides one potential
explanation for why some firms continue to operate with weaker governance when doing
so may mean lower credit ratings.
6.2. Endogeneity
The preceding analysis treats governance attributes as being exogenously determined.

Under the assumption of optimal contracting, a firm’s governance structure is unique in
equilibrium and endogenously determined (Bushman, Chen, Engel and Smith, 2004;
Hermalin and Weisbach, 2003). If governance provisions are endogenously determined
such that there is a factor or set of factors that affect governance and also affect credit
rating agencies’ assessment of firms’ creditworthiness, then our study suffers from a
potential correlated omitted variable problem. This misspecification causes the parameter
estimates to be inconsistent, which clouds the interpretation of results.
ARTICLE IN PRESS
Table 10
Relation between future performance, credit ratings and overcompensation
Panel A: Future performance and current credit rating (investment versus non-investment grade)
Dependent variable
Variables Predicted sign Next year’s ROA Next year’s ROE
ROA + 0.436***
ROE + 0.470***
INVESTMENT GRADE ? 0.009*** 0.055***
Adjusted R2 0.36 0.44
Sample size 869 869
Panel B: Future performance and current period overcompensation

Dependent variable
Variables Predicted sign Next year’s ROA Next year’s ROA Next year’s ROE
ROA + 0.498*** 0.480*** 0.496***
OC_Salary % 0.021***
OC Salary+Bonus% 0.024***
OC_Total Comp % 0.023***
2
Adjusted R 0.41 0.42 0.43
Notes: Panel A presents the coefficient estimates from the following OLS regressions:
Model 1: ROAtþ1 ¼ b0 þ b1 ROAt þ b2 INVESTMENT GRADEt þ ,
Model 2: ROEtþ1 ¼ b0 þ b1 ROEt þ b2 INVESTMENT GRADEt þ .
where ROE is equal to net income before extraordinary items (Compustat #18) divided by book value of common
equity (Compustat #60). See Tables 3 and 9 for other variable definitions. Panel B presents the coefficient
estimates from the following OLS regressions:
Model 1: ROAtþ1 ¼ b0 þ b1 ROAt þ b2 OC_Salary%t þ ,
Model 2: ROAtþ1 ¼ b0 þ b1 ROAt þ b2 OC_Salary þ Bonus%t þ ,
Model 3: ROAtþ1 ¼ b0 þ b1 ROAt þ b2 OC_TotalComp%t þ .
***,**,* indicates significance at the 0.01, 0.05, and 0.10 level or better, respectively. See Table 9 for variable
definitions.
The econometric solution for endogeneity is to use two-stage procedures that rely on
instrumental variables to generate predicted values of the independent variables (in our
case, the set of governance variables) that are uncorrelated with the error term in the
structural model. Unfortunately, instrumental variables are very difficult to identify in
most accounting research settings (Ittner and Larcker, 2001). This is particularly true with
respect to governance attributes in that there is no well-developed theory or model of the
economic determinants of governance (Hermalin and Weisbach, 2003).
The lack of theory on the determinants of corporate governance draws into question the
adequacy of any instrumental variable approach to deal with potential endogeneity
issues in our setting. However, there is limited empirical evidence (Hermalin and Weisbach,
ARTICLE IN PRESS
1991; Bhagat and Black, 2000), that poor past performance (both accounting and stock
market) leads to increases in board independence. Therefore, past performance is
potentially a correlated omitted variable, at least with respect to our board independence
measure.
In Table 11, we expand our full model (Model 6 in Table 6) to include stock returns
(PP_RET) as a past performance measure.24 This measure is industry-adjusted and we
present results for 1-, 3- and 5-year prior performance horizons.25 With one exception
(G_SCORE, 5-year prior performance), all of the variables that were significantly related
to credit ratings in our original model continue to be significant in the augmented model.
Thus, the inclusion of a past performance measure has little influence on our conclusions
regarding the importance of governance on credit ratings.
Another potential competing explanation for our results is that having debt in a firm’s
capital structure may act as a substitute for strong governance. Jensen (1986) argues that
having debt in the capital structure can reduce the agency costs that external stakeholders
face when firms have free cash flows available for discretionary spending by managers. The
threat caused by failure to make debt service payments serves as an effective motivating
force for managers to invest firms’ resources in ways that maximize the cash flows available
to external stakeholders. Firms with weak governance may choose to have managers pre-
commit to achieving a minimal cash flow by issuing debt. Thus, Jensen’s free cash flow
theory of capital structure implies that firms with weaker governance will have more debt
outstanding, resulting in lower credit ratings and higher interest costs.
Merging governance and the traditional capital structure hypothesis suggests that there
is a trade-off between governance and capital structure. In other words, strong governance
and leverage are negatively related. Thus, the higher interest costs we attribute to having
weaker governance may well be costs associated with firms choosing to operate with higher
leverage which serves the same role as governance in disciplining managements’
opportunistic tendencies towards over investing. However, there are several pieces of
evidence in our results that are inconsistent with this competing capital structure
hypothesis.
First, differences in sample firms’ capital structures are controlled for by including
leverage as one of the firm-specific control variables in our credit rating model. That is, the
effects of the various governance variables on credit ratings are after taking into account
how leverage affects credit ratings. Second, in untabulated results, we find the correlations
between the various governance variables and leverage to be quite low, generally falling in
the range of + or 0.05. Only one of the governance variables (%BRD_EXPERT)
exhibits a significant negative correlation with leverage (p ¼ 0:09). As a group, the
governance variables explain only 4% of the cross-sectional variation in leverage.
Therefore, for our sample, governance does not appear to determine leverage. Finally,
firms that issued debt during our sample period actually tended to have stronger
24
Including past performance measures in our base model along with the set of governance variable is
equivalent to using two-stage procedures where we first regress each of the governance variables on the past
performance variables and then include the predicted values from the first stage model into the structural model.
We choose the one-step approach because it is simpler to implement given the 14 separate governance variables in
our structural model.
25
We also conducted analyses using raw performance and market-adjusted performance measures and the
results are qualitatively the same as those reported here.
ARTICLE IN PRESS
Table 11
Sensitivity tests with prior period performance
Variables Predicted sign 1-Year prior 3-Year prior 5-Year prior

performance performance performance
LEV 2.154*** 1.794*** 1.763***
ROA + 0.107*** 0.089*** 0.095***
LOSS 1.089*** 1.155*** 1.045***
INT_COV + 0.015*** 0.029*** 0.025***
SIZE + 0.643*** 0.634*** 0.628***
SUBORD 0.569*** 0.544**** 0.580***
CAP_INTEN + 0.490*** 0.413** 0.244
FIN_UTILITY + 0.159 0.250 0.149
PP_RET ? 0.006*** 0.007** 0.009**
Ownership structure and
influence
BLOCK ? 0.156*** 0.169*** 0.176***
%INST ? 0.159 0.303 0.133
%INSIDE 0.497 0.461 0.503
Financial stakeholder rights and
relations
G_SCORE ? 0.040* 0.036* 0.032
WCAQ + 7.736*** 6.738*** 10.147***
TIMELINESS + 3.382*** 3.363*** 3.034****
TOTFEES 0.316 0.323 0.265
%AUD_IND + 0.063 0.216 0.637
FIN_EXPERT + 0.189 0.165 0.176
%BRD_IND + 1.107** 1.362** 1.337**
CEOPOWER 0.167 0.210* 0.245**
%BRD_EXPERT + 1.005*** 0.994** 1.214***
%BRD_STOCK + 0.915*** 1.127*** 1.133***
GOVERNANCE_POLICY + 0.116 0.179 0.191
%FINCOM_INSIDE + 0.276 0.713 0.952*
Generalized R2 0.60 0.60 0.60
Likelihood ratio w2 800.97*** 770.74*** 709.96***
Wald w2 125.40*** 128.03*** 117.07***
See Table 3 for variable definitions.Notes: Table values are coefficient estimates from the following ordered logit
models:
Model 1: RATING ¼ f (firm characteristics, prior 1-year return, governance attributes),
Model 2: RATING ¼ f (firm characteristics, prior 3-year return, governance attributes),
Model 3: RATING ¼ f (firm characteristics, prior 5-year return, governance attributes).
whether the governance attributes added in each model as a whole explain a significant portion of the variation in
firms’ credit ratings. RATING ¼ S&P Long Term Domestic Issuer Credit Rating (Compustat #280), see Table 1
for numeric coding, PP_RET is equal to the prior period(s) return over the fiscal year, in the 3 and 5 year columns
it is set to the average return over the past 3, 5 years. PP_RET is industry-adjusted, where industry groups are
defined by four, three, two and one digit SIC codes with a minimum of 10 firms in each industry group. See
Table 3 for other variable definitions.
ARTICLE IN PRESS
governance along most dimensions than those firms that did not issue debt.26 Again, these
results are inconsistent with debt acting as a substitute for strong governance.
In addition to the supplemental results noted above, there are other features of our
setting that suggest correlated omitted variables are not driving our results. In Table 6 we
show that there are seven distinct governance variables that are significantly related to
credit ratings. There is at least one variable from each of the four S&P components of
governance that exhibits significant explanatory power, and there is relatively low
correlation among these seven governance variables (see Table 5). Thus, there is no single
omitted variable that could simultaneously be correlated with all seven of these governance
variables in such a way to provide an alternative explanation for our results. Moreover, it
is hard to imagine that there would be a set of omitted economic variables that would be
highly correlated with our governance variables and be correlated with credit ratings in a
fashion that’s consistent with our findings.27
Another feature of our setting that suggests that we have appropriately modeled credit
ratings is the evidence from credit rating agencies themselves that indicates that
governance features are an important input into the credit rating process. For example,
in a recent special report on credit policy entitled, ‘‘Evaluating Corporate Governance:
The Bondholders’ Perspective’’28, FitchRatings (2004) states the following:
‘‘The purpose of this global criteria report is to inform the marketplace of Fitch
Ratings’ approach to evaluating and incorporating the quality of a company’s
corporate governance within the overall credit ratings process. While Fitch always
has taken aspects of corporate governance into account, this report formalizes a
more systematic framework for reviewing governance practices that affect credit
quality y’’ Fitch’s framework is grounded in agency theory and defines corporate
governance from a creditor perspectivey. Ultimately, companies that are found to
have exceptionally weak corporate governance (or disclosure practices) could face a
downgrade or other negative rating action, while those with very strong practices
might warrant a special or favorable mention in the credit analysis.’’ (p. 1)
Statements like these by major credit rating agencies clearly indicate that governance
factors are direct inputs to the credit rating process. Moreover, three major rating agencies
(S&P, Moody’s and Fitch Ratings) have developed infrastructures and have invested
significant resources to evaluate firms’ governance structures. These actions clearly signal
that governance is important to the credit rating process.
6.3. Sensitivity tests
In our original model, we use total audit fees paid to a firm’s auditor to proxy for the
economic bond between the auditor and client, which potentially threatens auditor
26
These results are available from the authors upon request.
27
We also acknowledge that there could be other economic variables omitted from the model that are correlated
(some positively and some negatively) with credit ratings. We have included all the major economic determinants
of credit ratings in our model based on evidence provided in prior research. If there are major economic variables
that have been omitted from our model, then these have been systematically overlooked by a vast literature on
determinants of debt ratings, and we believe this is unlikely.
28
Fitch Ratings, Credit Policy, Special Report, ‘‘Evaluating Corporate Governance: The Bondholders’
Perspective,’’ April 12, 2004.
ARTICLE IN PRESS
independence. Contemporaneous literature investigating the quality of accounting

information in the presence of threats to auditor independence uses alternative measures
of economic bonding (e.g., Frankel et al., 2002; Ashbaugh et al., 2003). To test the
robustness of our results related to the quality of the audit function in governance, we
substitute two alternative proxies of economic bonding for the TOTFEE variable in the
full model (not tabled). The first substitution uses the ratio of non-audit fees to total fees
(FEERATIO) for TOTFEE. The second substitution is a dummy variable coded one if
TOTFEE is in the upper quartile of the distribution of TOTFEE and zero otherwise
(FEEDUMMY). We expect observations falling into the upper quartile of TOTFEE to be
firms where auditor independence is more likely to be threatened. The coefficients on
FEERATIO and FEEDUMMY are insignificant at conventional levels. Thus, we continue
to find no evidence that measures of economic bonding between the auditor and client firm
adversely affects credit ratings.
Our second set of sensitivity tests relates to our proxy for transparent and timely
financial reporting. Recall that TIMELINESS is defined as negative one times the squared
residual from a cross-sectional regression of returns on the levels and changes in earnings.
We substitute two alternative measures of TIMELINESS in our full model. The first
substitution defines TIMELINESS as negative one times the variance of the residuals from
firm-specific time-series regressions, where we require firms to have a minimum
(maximum) of 8 (10) years of data to estimate Eq. (2). The coefficient on this specification
of TIMELINESS is positive and significant at the 0.001 level or better.
Gelb and Zarowin (2002) provide an alternative specification of the relation between
returns and earnings. Specifically, they posit and provide evidence that current price is
reflective of the informativeness of future earnings. Thus, our second substitution for
TIMELINESS is the negative squared residual from the regression of returns on
contemporaneous and future earnings and earnings changes after controlling for future
returns. Once again, the coefficient on this specification of TIMELINESS is positive and
significant at the 0.001 level or better.
We use the %BRD_EXPERT, the percentage of outside directors that sit on other boards,
as a measure of board competency or expertise following Klein (1998). However, there is
evidence in the literature that when board members sit on too many boards monitoring of
management is compromised and, as a consequence, firm performance deteriorates (Bhagat
and Black, 1999; Klein, 1998). As an additional sensitivity test (not tabled), we include a
variable for the percentage of board members that sit on four or more boards. We find no
evidence that board members being ‘‘too busy’’ adversely affects credit ratings, and adding this
variable does not detract from the significance of our %BRD_EXPERT variable.
The results of these sensitivity tests indicate that our inferences are robust to alternative
measures of governance attributes.
7. Summary and conclusions
Weak corporate governance has been singled out as the leading cause for recent high-
profile cases of corporate fraud. Using a framework for evaluating corporate governance
structures recently developed by Standard & Poor’s, this study investigates whether
firms that exhibit strong governance benefit from higher overall firm credit ratings rela-
tive to firms with weak governance. We present evidence that a variety of governance
attributes explain firm credit ratings after controlling for firm characteristics that prior
ARTICLE IN PRESS
research has shown to be related to credit ratings. Specifically, we find that firm credit
ratings are: (1) negatively associated with the number of blockholders that own at least a
5% ownership in the firm; (2) positively related to weaker shareholder rights in terms of
takeover defenses; (3) positively related to the degree of financial transparency; and (4)
positively related to overall board independence, board stock ownership and board
expertise and negatively related to CEO power on the board. We show that a hypothetical
firm that possesses desirable governance characteristics from the bondholder’s viewpoint
approximately doubles its likelihood of receiving an investment-grade credit rating. Given
the spread between investment-grade and speculative-grade bond yields, better governance
can translate into significant debt cost savings for firms.
Our primary analysis documents that firms’ governance affects firms’ credit ratings. Our
secondary analysis provides insights into why all firms do not possess strong governance.
We note that the cost of weak governance is borne by all stakeholders whereas the benefits
of weak governance can accrue to managers when they can appropriate some or all of the
rents from outside stakeholders by resisting better governance. Our findings suggest that
CEOs of weak governance firms garner overcompensation in excess of their share of
additional debt costs due to weak governance. Thus, we provide one explanation for why
all firms do not practice good governance.
A number of organizations and companies (Standard & Poor’s, Board Analyst, The Board
Institute, Moody’s Investor Services, and FitchRatings) have begun to compile company
ratings of corporate governance practices along several dimensions. Investigating whether
these composite ratings are useful determinants of credit ratings is one avenue of future
research. Another avenue of future research is to focus on the benefits of governance to equity
stakeholders by investigating the relation between governance and firms’ cost of equity capital.
Appendix
We use the following alternative measures of CEO pay (see Core et al., 1999; Aboody
et al., 2004) to derive measures of overcompensation:
Salary ¼ the dollar value ($000) of the base salary (cash and non-cash) earned by the CEO
during fiscal 2002,
Salary+Bonus ¼ current compensation ($000) comprised of salary and bonus earned by
the CEO during fiscal 2002,
TotalComp ¼ total compensation ($000) earned by the CEO during fiscal 2002, comprised
of the following: salary, bonus, other annual, total value of restricted stock granted, total
value of stock options granted (using Black-Scholes), long-term incentive payouts, and all
other total.
Following Core et al. (1999), we use the following variables to measure the economic
determinants of CEO pay:
Sales ¼ the natural log of Compustat #12,
MB ¼ the average market to book ratio as of the end of fiscal 2001 (Compustat #25 x #199
divided by #60) where firms are required to have a minimum of 3 and a maximum of 5
years of prior data,
OPROA ¼ operating income divided by total assets (Compustat #13 minus #14 divided
#6) for the 2001 fiscal year,
ARTICLE IN PRESS
RET ¼ the buy and hold return over the 2001 fiscal year,
STD_OPROA ¼ the standard deviation of OPROA as of the end of the 2001 fiscal year
where firms are required to have a minimum of 3 and a maximum of 5 years of OPROA,
STD_RET ¼ the standard deviation of buy and hold returns for the fiscal year where firms
are required to have a minimum of 3 and a maximum of 5 years of RET.
We use the following board structure and ownership structure variables to capture CEO
power/management entrenchment (the predicted relation with CEO compensation is
shown in parentheses—see Core et al. (1999) and previous discussion in this paper):
CEOPOWER ¼ Composite score representing the power of the CEO where a firm receives
one point if the CEO is the chairman of the board, and one point for each of the
committees (compensation, nominating, audit) that the CEO sits on (in some instances the
CEO does not have voting power yet still is identified as being part of the committee)
(source Board Analyst) (+),
%COMP_CEOAPP ¼ the percent of outside independent directors on the compensation
committee appointed by the CEO (+),
BRDSIZE ¼ the number of directors on the board (+),
%BRD_INSIDE ¼ the percent of the board made up of insiders (?),
%OUT_BUSY ¼ the percent of outside independent directors that sit on more than four
boards (number of outside independent directors sitting on more than four boards divided
by total outside independent directors) (+),
%BRD_ActiveCEOs ¼ the percent of the board that are currently employed as CEO’s (+),
%INST ¼ % of shares held by institutional investors multiplied by 100 (source Board
Analyst) (-),
BLOCK ¼ Number of blockholders, where block is defined at the 5% ownership level
(source Compact Disclosure) (),
IND_DUM ¼ industry dummies based on firms’ two-digit SIC codes and are included for
each two-digit SIC group having at least 10 observations, for a total of 23 industry
dummies.
The following cross-sectional OLS model is used to estimate the determinants of CEO
compensation using fiscal year 2002 data for our sample firms:
X
23
CEO_COMPi ¼ b0n IND_DUMni þ b1 Salesi þ b2 MBi þ b3 OPROAi þ b4 RETi
n¼1
þ b5 STD_OPROAi þ b6 STD_RETi þ b7 CEOPOWERi
þ b8 %COMP_CEOAPPi þ b9 BRDSIZEi þ b10 %BRD_INSIDEi
þ b11 %OUT_BUSYi þ b12 %BRD_ActiveCEOsi
þ b13 CEO_Ownershipi þ b14 %INSTi þ b15 BLOCKi þ i . ðA1Þ
References
Aboody, D., Barth, M., Kasznik, R., 2004. Do firms manage stock-based compensation expense disclosed under
SFAS 123? Working paper, University of California at Los Angeles and Stanford University.
Agrawal, A., Chadha, S., 2003. Corporate governance and accounting scandals. Working paper, University of
Alabama.
ARTICLE IN PRESS
Agrawal, A., Knoeber, C., 1996. Firm performance and mechanisms to control agency problems between
managers and shareholders. Journal of Financial and Quantitative Analysis 31, 377–397.
Agresti, A., 2002. Categorical Data Analysis. John Wiley and Sons, New York.
Ashbaugh, H., LaFond, R., Mayhew, B., 2003. Do nonaudit services compromise auditor independence? Further
evidence. Accounting Review 78, 611–640.
Asquith, P., Wizman, T., 1990. Event risk, covenants, and bondholder returns in leveraged buyouts. Journal of
Financial Economics 27, 195–213.
Barth, M., Landsman, W., 2003. Cost of capital and the quality of financial statement information. Working
paper, Stanford University.
Baysinger, B., Butler, H., 1985. Corporate governance and the board of directors: performance effects of changes
in board composition. Journal of Law, Economics and Organizations 1, 101–124.
Bhagat, S., Black, B., 1999. The uncertain relationship between board composition and firm performance.
Business Lawyer 54, 921–963.
Bhagat, S., Black, B., 2000. Board independence and long-term performance. Working paper, Stanford Law
School, Stanford, CA.
Bhojraj, S., Sengupta, P., 2003. Effect of corporate governance on bond ratings and yields: the role of institutional
investors and the outside directors. The Journal of Business 76, 455–475.
Boardman, C., McEnally, R., 1981. Factors affecting seasoned corporate bond prices. Journal of Financial and
Quantitative Analysis 16, 207–216.
Bushman, R., Chen, Q., Engel, E., Smith, A., 2004. Financial accounting information, organizational complexity
and corporate governance systems. Journal of Accounting & Economics 37, 167–202.
Chung, H., Kallapur, S., 2003. Client importance, nonaudit services, and abnormal accruals. Accounting Review
78, 931–955.
Core, J., Holthausen, R., Larcker, D., 1999. Corporate governance, chief executive officer compensation, and firm
performance. Journal of Financial Economics 51, 371–406.
Dann, L., DeAngelo, H., 1983. Standstill agreements, privately negotiated stock repurchases, and the market for
corporate control. Journal of Financial Economics 11, 275–300.
Davidson, W., Xie, B., Xu, W., 2004. Market reaction to voluntary announcements of audit committee
appointments: the effect of financial expertise. Journal of Accounting & Public Policy 23, 279–293.
DeAngelo, L., 1981. Auditor size and audit quality. Journal of Accounting & Economics 3, 183–200.
Dechow, P., Dichev, I., 2002. The quality of accruals and earnings: the role of accrual estimation errors.
Accounting Review 77, 35–59.
DeFond, M., Raghunandan, K., Subramanyam, K.R., 2002. Do non-audit service fees impair auditor
independence? Evidence from going concern audit opinions. Journal of Accounting Research 40, 1247–1274.
DeFond, M., Hann, R., Hu, X., 2004. Does the market value financial expertise on the audit committee of boards
of directors? Journal of Accounting Research (Supplement) 43, 153–194.
FitchRatings, 2004. Credit policy special report, evaluating corporate governance: the bondholders’ perspective,
New York.
Frankel, R., Johnson, M., Nelson, K., 2002. The relation between auditors’ fees for nonaudit services and
earnings management. Accounting Review 77, 71–105.
Gelb, D., Zarowin, P., 2002. Corporate disclosure policy and the informativeness of stock prices. Review of
Accounting Studies 7, 33–52.
Gompers, P., Ishii, J., Metrick, A., 2003. Corporate governance and equity prices. Quarterly Journal of
Economics 118, 107–155.
Gordon, L., Pound, J., 1993. Information, ownership structure, and shareholder voting: evidence from
shareholder-sponsored corporate governance proposal. Journal of Finance 48, 697–718.
Grinblatt, M., Titman, S., 2002. Financial Markets and Corporate Strategy, second ed. McGraw-Hill Irwin, New
York.
Gu, Z., 2002. Cross-sample incomparability of R2s and additional evidence on value relevance changes over time.
Working paper, Carnegie Mellon University.
Hermalin, B., Weisbach, M., 1991. The effect of board composition and direct incentives on firm performance.
Financial Management 21, 101–112.
Hermalin, B., Weisbach, M., 2003. Boards of directors as an endogenously determined institution: a survey of the
economic literature. Economic Policy Review 9, 7–26.
Horrigan, J., 1966. The determinants of long-term credit standing with financial ratios. Journal of Accounting
Research 4, 44–62.
ARTICLE IN PRESS
Imhoff, E.A., 2003. Accounting quality, auditing, and corporate governance. Accounting Horizons Supplement
17, 117–128.
Ittner, C., Larcker, D., 2001. Assessing empirical research in managerial accounting: a value-based management
perspective. Journal of Accounting & Economics 32, 349–410.
Jensen, M., 1993. The modern industrial revolution, exit, and the failure of internal control systems. Journal of
Finance 48, 831–880.
Jensen, M., Meckling, W., 1976. Theory of the firm: managerial behavior, agency costs and ownership structure.
Journal of Financial Economics 3, 305–360.
Kaplan, R., Urwitz, G., 1979. Statistical models of bond ratings: a methodological inquiry. Journal of Business
52, 231–261.
Karpoff, J., Malatesta, P., Walkling, R., 1996. Corporate governance and shareholder initiatives: empirical
evidence. Journal of Financial Economics 42, 365–395.
Klein, A., 1998. Firm performance and board committee structure. Journal of Law and Economics 41, 275–303.
Klein, A., 2002. Audit committee, board of director characteristics, and earnings management. Journal of
Accounting & Economics 33, 375–400.
Klein, A., 2003. Do audit committees with financially literate directors experience less earnings management.
Working paper, Stern School of Business, New York University.
Klock, M., Mansi, S., Maxwell, W., 2004. Corporate governance and the agency cost of debt. Working paper,
George Washington University.
Lamy, R., Thompson, R., 1988. Risk premia and the pricing of primary issue bonds. Journal of Banking and
Finance 12, 585–601.
McConnell, J., Serves, H., 1990. Additional evidence on equity ownership and corporate value. Journal of
Financial Economics 27, 595–612.
Nesbitt, S., 1994. Long-term rewards from shareholder activism: a study of the ‘CalPERS’ effect. Journal of
Applied Corporate Finance 6, 75–80.
Opler, T., Sokobin, J., 1997. Does coordinated institutional activism work? An analysis of the activities of the
council of institutional investors. Working paper, Ohio State University.
Securities and Exchange Commission, 2003, NASD and NYSE rulemaking: relating to corporate governance.
Release No. 34-48745, Government Printing Office, Washington, DC.
Sengupta, P., 1998. Corporate disclosure quality and the cost of debt. Accounting Review 73, 459–474.
Shleifer, A., Vishny, R., 1997. A survey of corporate governance. Journal of Finance 52, 737–783.
Standard & Poor’s, 2002. Standard & Poor’s Corporate Governance Scores: Criteria, Methodology and
Definitions. McGraw-Hill Companies, Inc., New York.
U.S. Census Bureau, 2005. Statistical Abstract of the United States: 2004–2005, Section 25 Banking, Finance and
Insurance. Government Printing Office, Washington, DC.
U.S. Congress, 2002. The Sarbanes-Oxley Act of 2002, 107th Congress of the United States of America. H.R.
3763. Government Printing Office, Washington, DC.
Warga, A., Welch, I., 1993. Bondholder losses in leveraged buyouts. Review of Financial Studies 6, 959–982.
Yermack, D., 1996. Higher market valuation of companies with a small board of directors. Journal of Financial
Economics 40, 185–211.
Yermack, D., 2003. Remuneration, retention, and reputation incentives for outside directors. Journal of Finance
59, 2281–2308.
Ziebart, D., Reiter, S., 1992. Bond ratings, bond yields and financial information. Contemporary Accounting
Research 9, 252–282.

Effect of Corporate Governance On Credit Ratings

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Effect of Corporate Governance On Credit Ratings

Uploaded by

Copyright:

Available Formats

ARTICLE IN PRESS

Journal of Accounting and Economics 42 (2006) 203–243

The effects of corporate governance on ﬁrms’

JEL classification: G30; G32; M41

Keywords: Corporate governance; Credit rating; Executive compensation

2. Why governance affects ﬁrms’ credit ratings

perspective. We refer to this as the ‘‘wealth redistribution’’ hypothesis regarding how

3. Corporate governance attributes

3.1. Ownership structure and influence

3.2. Financial stakeholder rights and relations

Financial stakeholder rights reﬂect the balance of power between stakeholders

3.3. Financial transparency

Transparent ﬁnancial reporting is critical to reducing the information asymmetry

3.4. Board structure and processes

4. Sample, variables, and descriptive statistics

4.1. Sample and data sources

Data for this study are compiled from four sources:

Governance measures, audit/non-audit fees. and share ownership data—Board Analyst

S&P Debt rating Compustat data280 Assigned RATING Grade

Panel A: Sample construction (fiscal year 2002)

Number of ﬁrms in the main Board analyst dataset 2050 0

Panel B: Sample firms’ board composition

Number of ﬁrms having a 894 894 890 813 242

4.2. Independent variables

4.2.1. Corporate governance measures

Variables Predicted Deﬁnitions and data source

Ownership structure and inﬂuence

Financial stakeholder rights and relations.

Board structure and processes

Variables Predicted Deﬁnitions and data source

%FINCOM_INSIDE + % of insiders on the ﬁnance committee (source proxy statement).

Our ﬁrst measure of ﬁnancial transparency, WCAQ, is an accounting-based measure of

4.2.2. Control variables—firm characteristics

The speciﬁc measurements of the variables capturing differences in ﬁrm characteristic

4.2.3. Descriptive statistics

Variables Mean Standard Median 25% 75%

Ownership structure and inﬂuence

directors) is 8% (4%). For the ‘‘Financial Stakeholder Rights’’ dimension of corporate

Panel A: Firm characteristics

RATING 0.28 0.44 0.42 0.27 0.50 0.19 0.03 0.24

Panel B: Governance variables

coding. See Table 3 for other variable deﬁnitions.

5. Empirical tests and results

5.1. Ordered logit results

Variables Predicted Estimated coefﬁcient

Sign Model 1 Model 2 Model 3

Financial transparency and information disclosure

characteristics (generalized R2 of 0.41 versus 0.52). In summary, we ﬁnd that governance

5.2. Investment versus speculative-grade analysis

Variables Predicted sign Estimated coefﬁcient

An alternative way of assessing the effect of various governance variables on the

Variables Predicted sign Marginal effect Change in probability

6.1. Why don’t all firms practice good governance?

CEO_PAY ¼ f ðeconomic determinants; board and ownership structure attributesÞ.

Panel A: OLS model of CEO compensation

Salary Salary+Bonus Total compensation

Panel B: Estimates of overcompensation by grade

OC_Salary % (mean) 0.349 0.391 0.00

Panel C: Overcompensation to share of interest costs ratio—speculative grade firms (n ¼ 245)

OC_Salary/CEOSHARE_INTEXP 51.242 122.816 435.501

OC_Salary/CEOSHARE_INTEXP 5 240 225.41***

The preceding analysis treats governance attributes as being exogenously determined.

Variables Predicted sign Next year’s ROA Next year’s ROE